Group Title: assessment of the efficiency and effectiveness of simulated auto racing performance
Title: An assessment of the efficiency and effectiveness of simulated auto racing performance
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00100812/00001
 Material Information
Title: An assessment of the efficiency and effectiveness of simulated auto racing performance psychophysiological evidence for the processing efficiency theory as indexed through visual search characteristics and P300 reciprocity
Physical Description: Book
Language: English
Creator: Murray, Nicholas P., 1969-
Publisher: State University System of Florida
Place of Publication: Florida
Florida
Publication Date: 2000
Copyright Date: 2000
 Subjects
Subject: Exercise and Sport Sciences thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Exercise and Sport Sciences -- UF   ( lcsh )
Genre: government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )
 Notes
Summary: ABSTRACT: This study was undertaken to examine the central tenets of the processing efficiency theory (PET) with more direct measures of attention and effort. Participants were placed into either higher and lower trait anxiety groups and required to concurrently perform a simulated driving task while responding to one of four target LED's (which could only be differentiated by a goal directed saccade) upon presentation of a valid or invalid cue located in the central or peripheral visual field. Eye movements, cortical activity and dual task performance were recorded under two conditions: baseline, and competition, where cognitive anxiety was induced by an instructional set. Findings indicated that, while there was little change in driving performance from Session 1 to Session 2, response time was reduced for the lower anxious group with an increase for the higher anxious group in the competitive session. In addition, search rate was increased and a reduction in P3 amplitude occurred for both groups during the competitive session, which revealed a reduction in processing efficiency as indexed by state anxiety. Implications of this study provide a more comprehensive and complex account of the theory, and suggest that increases in cognitive anxiety result in a reduction of processing efficiency with little change in performance effectiveness.
Summary: KEYWORDS: anxiety, attention, processing efficiency
Thesis: Thesis (Ph. D.)--University of Florida, 2000.
Bibliography: Includes bibliographical references (p. 106-117).
System Details: System requirements: World Wide Web browser and PDF reader.
System Details: Mode of access: World Wide Web.
Statement of Responsibility: by Nicholas P. Murray.
General Note: Title from first page of PDF file.
General Note: Document formatted into pages; contains vi, 124 p.; also contains graphics.
General Note: Vita.
 Record Information
Bibliographic ID: UF00100812
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 47682152
alephbibnum - 002678734
notis - ANE5961

Downloads

This item has the following downloads:

Dissertation_Nicholas_Murray ( PDF )


Full Text











AN ASSESSMENT OF THE EFFICIENCY AND EFFECTIVENESS OF SIMULATED
AUTO RACING PERFORMANCE: PSYCHOPHYSIOLOGICAL EVIDENCE FOR
THE PROCESSING EFFICIENCY THEORY AS INDEXED THROUGH VISUAL
SEARCH CHARACTERISTICS AND P300 RECIPROCITY

















By

NICHOLAS P. MURRAY


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA
2000














ACKNOWLEDGEMENTS

I am grateful to my advisor, Dr. Christopher Janelle, for his support and guidance

throughout my doctoral education at the University of Florida. I have benefited by his

stimulating approach to research and his invaluable help in preparation of this document.

Secondly, I thank my committee members Dr. James Cauraugh, Dr. Ira Fischler, and

Dr. Robert Singer for their supportive effort and additional insight into my work. I

would also like to thank Dr. Keith Tennant for his help in the beginning of my doctoral

education and his continual support. I thank Curtis Weldon for his expertise, and his

help in developing the stimulus/response apparatus employed in my dissertation data

collection.

Special thanks go to my wife, Melissa Murray, for the happiness shared, her

patience, and loving support. I am particularly indebted to her for the encouragement that

guided me through the completion of this dissertation. I am forever grateful to my

parents, Dr. Thomas Murray and Dr. Margaret Murray, who have also given insight,

direction and support without this I could not have accomplished so much.















TABLE OF CONTENTS
Page

ACKNOWLEDGEMENTS ii

ABSTRACT v

CHAPTERS
1 INTRODUCTION 1
Statement of the Problem 3
Hypotheses - 4
Significance of the Study ......................................................8..... 8

2 REVIEW OF LITERATURE 10
2 REV EW OF LITERATURE...................................................10

A rousal, A nxiety, and Stress.......................................................10
Anxiety and Perform ance . ---------------------...... 13
Facilitating and Debilitating Effects of Anxiety.........................14
Theories of Anxiety .. . . . . . . . 16
Reaction Time and Orienting of Attention .......... ..................... 33
Anxiety, Visual Search, and Visual Attention _-----44
Event Related Potentials (ERP) and Mental Effort ....................49
Summary 62

3 METHODS 64

P participants .. .. .. .. .. .. .. .. .. . 64
Tasks, Instruments, and Objectives ............................................65
Measurement Devices and Performance Measures 67
Procedure 71
Design and Analysis -----------------------------74

4 RESULTS 76

Pretest Performance Criterion 76
T rait A nxiety ............................................................................. . 76
Cognitive Anxiety and Physiological Arousal ------------------77
Dual-task Performance 78
Visual Search Data 80
ER P D ata A analysis .......................................................................85









5 DISCUSSION, SUMMARY, CONCLUSIONS, AND
SUGGESTIONS FOR FUTURE RESEARCH 88

Discussion 89
Summary 100
Conclusions and Im plications ....................................................103
Future Directions 104

REFERENCES 106

APPENDICES

A COMPETITIVE STATE ANXIETY INVENTORY 2
(C SA I-2) ...................................................................................... 118

B VIDEO GAME QUESTIONNAIRE ................................. 119

C INFORMED CONSENT FORM 120

D EXPERIMENTAL MANIPULATION 122

BIOGRAPHICAL SKETCH 124














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

AN ASSESSMENT OF THE EFFICIENCY AND EFFECTIVENESS OF SIMULATED
AUTO RACING PERFORMANCE: PSYCHOPHYSIOLOGICAL EVIDENCE FOR
THE PROCESSING EFFICIENCY THEORY AS INDEXED THROUGH VISUAL
SEARCH CHARACTERISTICS AND P300 RECIPROCITY


By

Nicholas P. Murray

December 2000


Chair: Dr. Christopher Janelle
Major Department: Exercise and Sport Sciences

This study was undertaken to examine the central tenets of the processing

efficiency theory (PET) with more direct measures of attention and effort. Participants

were placed into either higher and lower trait anxiety groups and required to concurrently

perform a simulated driving task while responding to one of four target LED's (which

could only be differentiated by a goal directed saccade) upon presentation of a valid or

invalid cue located in the central or peripheral visual field. Eye movements, cortical

activity and dual task performance were recorded under two conditions: baseline, and

competition, where cognitive anxiety was induced by an instructional set. Findings

indicated that, while there was little change in driving performance from Session 1 to

Session 2, response time was reduced for the lower anxious group with an increase for

the higher anxious group in the competitive session. In addition, search rate was









increased and a reduction in P3 amplitude occurred for both groups during the

competitive session, which revealed a reduction in processing efficiency as indexed by

state anxiety. Implications of this study provide a more comprehensive and complex

account of the theory, and suggest that increases in cognitive anxiety result in a reduction

of processing efficiency with little change in performance effectiveness.














CHAPTER 1
INTRODUCTION

One of the ultimate goals of sport psychologists and coaches alike is to help

athletes achieve peak performance. For athletes, the ability to perform at the highest level

possible, which includes achieving an appropriate mental state prior to and during

performance, is paramount. However, this goal is fraught with the frailties of mental

control and the intrusion of thought processes, emotions, and situational factors that

frequently undermine the attainment of peak performance. Highly competitive and

demanding situations will often produce increased stress, heightened arousal, and

increased anxiety, especially in situations where the outcome is important and uncertain

(Jones & Swain, 1995). Not surprising is the substantial proportion of time sport

psychologists spend helping athletes to cope with the stress and anxiety that often occurs

prior to and during athletic competition. With this in mind, sport psychology researchers

have devoted a considerable amount of effort to understanding the anxiety-performance

relationship. While research on anxiety is extensive, understanding of the anxiety-

performance relationship is still somewhat limited.

Several conceptual frameworks have been advanced to address the

stress/performance relationship. These include drive theory (Hull, 1943; Spence &

Spence, 1966), the inverted-U hypothesis (Yerkes & Dodson, 1908), the

multidimensional anxiety theory (Martens, Vealey, & Burton, 1990), the zone of optimal

function (Hanin, 1980), the cusp catastrophe model of anxiety (Hardy & Fazey, 1987),

and the processing efficiency theory (PET) (Eysenck & Calvo, 1992). While each of









these concepts has strengths and weaknesses, only the PET and the cusp catastrophe

model provide an explanation that addresses both the debilitating and facilitating effects

of anxiety. In early anxiety research, anxiety was considered detrimental to performance;

however a growing body of literature suggests that anxiety has both facilitative and

debilitative functions (e.g. Jones, Hanton, & Swain, 1994; Jones & Swain, 1992; Jones,

Swain, & Hardy, 1993). Furthermore, the processing efficiency theory (Eysenck &

Calvo, 1992) provides an explanation of how individual differences in trait anxiety affect

performance, and how attentional mechanisms influence anxiety. Thus, the PET is an

ideal theory to explain the changes in motor performance as it relates to anxiety.

According to the PET, the level of performance is determined by state anxiety,

which is a product of the interaction between trait anxiety and situational stress (Eysenck

& Calvo, 1992). Worry, or cognitive anxiety, absorbs available resources from the

working memory system (Baddeley, 1986). The presence of cognitive anxiety produces a

motivational property, which will lead to the allocation of additional processing resources

for the task. The goal of increasing resource allocation is to prevent cognitive anxiety

from impairing task performance, and to reduce the impact of increased cognitive

anxiety. As such, predicted is that high trait-anxious individuals will exhibit a greater

probability to allocate additional resources, or effort to task performance than low trait-

anxious individuals. Thus, a crucial theoretical distinction is made between performance

effectiveness and processing efficiency. Performance effectiveness is the quality of the

performance, whereas processing efficiency is the relationship of performance

effectiveness and amount of cognitive effort invested in the task. This distinction is

fundamental to the theory, with the effects of anxiety on effectiveness and efficiency of









performance being quite different. To test the theory, effectiveness and efficiency must

be measured separately, which is generally accomplished through a dual task paradigm.

Currently, support for the PET is rather simplistic, and does not directly account

for the proposed underlying mechanisms such as the impact of anxiety on the working

memory system, and the reallocation of processing resources to the primary task. The

theoretical assumptions of the PET have been tested in relatively indirect fashion, and

surmised from tasks that contain substantial verbal components (e.g., Dibartolo, Brown,

& Barlow, 1997). While much of the research has inferred the processing efficiency and

performance effectiveness distinction in dual task situations, little empirical evidence has

been provided to directly confirm the allocation of additional processing resources by

anxious individuals. In effort to understand the reallocation of processing resources, a

more comprehensive and complex account of the theory is necessary. To this end, the

inclusion of more sophisticated measures to demonstrate the reallocation of attention is

needed. Specifically in this study, visual search and psychophysiological measures were

employed to delineate the processing efficiency and performance effective distinction, as

well as to provide more direct evidence for the reallocation of attention in a stressful

situation. Thus, this project attempted to extend previous research by examining the

processing changes of higher, and lower anxious participants through an "online"

method, and through a dynamic and complex motor task.

Statement of the Problem

The primary purpose of this study was to test the effectiveness and efficiency

distinction using a high speed-driving simulator, and, dependent on outcome, either offer

support for its use in sport psychology research or explain how the theory may be

extended to better describe the conditions under which anxiety might be either beneficial









or detrimental to performance. Thus, an assessment was made of whether changes in

performance on the primary task, response time needed to perform the secondary task,

visual search patterns, and cortical activation represent a reduction in processing

efficiency for higher anxious performers when compared to lower anxious performers.

The secondary purpose was to further knowledge related to the interaction of attentional

allocation, cortical activity and the ocular system in reactive settings. The available

evidence (e.g., Hoffman & Subramaniam, 1995) suggests that attention allocation

precedes saccadic shifts of attention. Therefore, changes in attentional allocation will

impact both cortical activity and visual search patterns. Specifically, the secondary

purpose is to evaluate the selective allocation of limited attentional resources through an

overt shift of attention as indexed by visual search and ERP data.

Using a racecar-driving simulator in conjunction with a cost and benefit

paradigm, the processing efficiency theory and attention allocation were analyzed.

Processing efficiency and attention allocation were tested during two sessions, baseline

and competition, by inducing anxiety through the use of an anxiety instructional set

similar to that used by Janelle, Singer, and Williams (1999). Based on pretest measures

of anxiety, participants were assigned to either a higher anxiety group or a lower anxiety

group. Dependent variables included: (1) driving speed, (2) response time to peripheral

lights, (3) visual search patterns, (4) mental effort as assessed by event-related brain

potentials, and (5) anxiety levels assessed by the CSAI-2.









Hypotheses

To address the issues presented above the following hypotheses were tested in this

investigation. The first set of hypotheses addresses the implications of the dual tasks and

the introduction of anxiety through an anxiety instructional set.

1. The anxiety instructional set was predicted to produce higher levels of

cognitive anxiety as measured by the CSAI-2 in the competition session as compared to

the baseline session. The competitive session was also hypothesized to produce higher

cognitive anxiety for higher trait anxious participants when compared to lower trait

anxious individuals. The anxiety instructional set has been shown to be a valid and

reliable method to induce changes in anxiety associated with competitive situations

(Hardy et al., 1994; Janelle et al., 1999).

2. The increase in anxiety in the competition session was expected to generate a

corresponding increase in physiological arousal (as measured by heart rate) in both

groups. The increase was predicted to be highest immediately prior to the start of the

competitive session (e.g., Swain & Jones, 1992).

3. Anxiety was hypothesized to impair processing efficiency to a greater extent

than it would impair performance effectiveness for higher trait anxious individuals

relative to lower trait anxious individuals (Eysenck & Calvo, 1992). It was expected that

these groups would only differ slightly on performance of the primary task while

significantly differing in performance of the secondary task. Specifically, higher trait

anxious individuals would have significantly longer response times than lower trait

anxious individuals with regard to performance in the secondary task (e.g., Dibartolo,









Brown, & Barlow, 1997), as a result of increased allocation of attentional resources to

maintenance of performance in the primary task.

The second set of hypotheses addresses the cost and benefit paradigm, response

time, and the processing efficiency theory, and are generated based previous findings of

the cost and benefit paradigm in which an individual will have slower response times

when anticipating or cued incorrectly, and faster response times when anticipating or

cued correctly.

1) In the single task trial and the non-competitive session, it was hypothesized

that valid peripheral cues would produce the fastest response times followed by valid

central cues, invalid central cues, and invalid peripheral cues, which were expected to

produce the slowest response times (Jonides, 1981).

2) Response times were predicted to increase for valid and invalid centrally

located cues in the competitive session for higher trait anxious individuals whereas when

compared to the non-competitive session. Central cues were considered voluntary, more

resource demanding, and more vulnerable to disruption than peripheral cues, which were

considered involuntary, strongly automatic, and resistant to disruption (Jonides, 1981).

Higher trait anxious individuals were expected to increase effort in the primary task, thus

decreasing the allocation of resources to the secondary task. Lower trait anxious

individuals were predicted to have more resources available and therefore be able to more

efficiently allocate resources to the central cues leading to faster response times.

3) Because peripheral cues were considered resistant to attentional disruption

and required few cognitive resources, it was expected that valid and invalid peripheral

cues would not significantly differ from non-competitive to competitive sessions.









The following hypotheses addressed the changes in visual search patterns and P3

amplitude.

1. Impairments in processing efficiency due to the presence of anxiety were

hypothesized to result in an increase in search rate. Specifically, high trait anxious

individuals were expected to produce more fixations for shorter durations. This was

considered a critical index of search efficiency, and if supported would indicate that

higher anxious individuals required more fixations to acquire the same information as

was previously gained with less fixations (under lower anxiety levels). Because it was

hypothesized that higher trait anxious individuals would apply more resources to the

central task, they would receive less information to perform the secondary task well, and

thus require more saccadic activity to peripheral locations to achieve the desired

secondary task goals.

2. Following the same logic from the previous hypothesis, it was expected that

higher anxiety would cause an increase in the number of fixations to peripheral locations

(Janelle et al., 1999).

3. The modified cost and benefits paradigm was hypothesized to generate an

attentional enhancement and a decrease in saccade latency when the cued location and

the stimulus presentation location were the same. Conversely, when the cued location

and the stimulus presentation location were different, interference was expected to

emerge, similar to a dual task situation. This would be evidenced by an increase in

saccade latency as well as an increase in response time (Hoffman & Subramaniam,

1995).









4. According to the PET, highly anxious performers were expected to apply

more resources to the primary task and have fewer resources available for secondary task

performance. P3 amplitude reciprocity is considered to be a measure of the degree to

which an individual is attending to either task in a dual task paradigm (Nash &

Fernandez, 1996). It was therefore hypothesized that as higher trait anxious individuals

apply more effort to the primary task, the P3 amplitude for the secondary task would

decrease (Wickens et al., 1983). A decrease of 5 microvolts to 8 microvolts was

expected, which would represent a statistically significant reduction in amplitude in

Session 2 when compared to Session 1.

Significance of the Study

This study presented a unique approach to understanding the anxiety-performance

relationship, and offered a more comprehensive account of the PET. The critical

question asked was whether anxiety has an influence on cognitive effort. If it did affect

effort, a follow-up question was: Does the PET adequately account for the anxiety-

performance relationship? To this end, a primary goal of this study was to further

substantiate the proposed mechanisms underlying the PET. Secondly, the viability of the

PET as a conceptual framework for sport anxiety research was examined.

The information gathered in this study also extended knowledge concerning the

understanding of competitive-anxiety during the event itself, as well as provided more

detailed evidence concerning psychophysiological evidence for the PET. The results are

important not only for researchers who study anxiety and performance, but also sport

psychologists, coaches, and athletes who may eventually use the information to improve

mental strategies, which could increase performance effectiveness. By offering some

insight to the underlying mechanisms that influence performance, and based on the some






9


limitations of previous research, this study offers a more complex and direct account of

the anxiety-performance relationship. Implication is the creation of intervention

protocols to increase processing efficiency and eventual performance effectiveness.














CHAPTER 2
REVIEW OF LITERATURE

This chapter presents a review of literature associated with theories of anxiety,

with a particular focus on the processing efficiency theory. In an effort to offer a

comprehensive overview of anxiety and prior to discussing the processing efficiency

theory, the first section provides an overview of the concepts and definitions that are

typically used to explain the anxiety-performance relationship. In this discussion, a

distinction is made between arousal, stress, and anxiety, as well as an elaboration of the

debilitating and facilitating effects of anxiety. The second section contains a delineation

and critique of the above theories, and a rationalization for the PET. In the third section,

testing the predictions of the PET using a probe task, visual search, and

electroencephalogram (EEG) is presented. A brief summary is outlined in the final

section.

Arousal, Anxiety, and Stress

It is frequently difficult to separate interpretation of the terms, arousal, stress, and

anxiety, because they are often used interchangeably. However, each are differentiated

by their origin, and by how they are manifested. Arousal is a physiological response to

environmental stimuli whereas stress and anxiety represent the emotional and cognitive

impact of arousal. The following paragraphs are included to better define and distinguish

these concepts.









Arousal

Arousal, according to Posner and Boies (1971), is one of the three important

components of attention, with limited information processing capacity and selective

attention being the other two. Arousal is generally considered synonymous with the

notion of alertness and physiological level of activation that varies on a continuum from

deep sleep to extreme excitement (Martens, Vealey, & Burton, 1990).

Neurophysiologically, arousal is the degree of activation of the organs and

mechanisms that are under the control of the body's autonomic nervous system (ANS)

(Guyton & Hall, 1996). The ANS is directly related to arousal and activation, and

controls the nerves in the smooth muscles and glands of the body. The ANS is divided

into the sympathetic and parasympathetic divisions. The sympathetic division, which is

primarily responsible for changes in bodily function associated with arousal, causes

increased sweating of the palms, increased heart rate, pupil dilation, increased respiration,

increased muscle tension, release of glucose from the liver, and decreased kidney output.

The parasympathetic division reduces the effects of the sympathetic division and initiates

a return to homeostasis. The general observation by researchers is that organisms seek to

attain an optimal level of arousal that is ideal for the task at hand (Lacey & Lacey, 1958).

Anxiety

Simply defined, anxiety is subjective feelings of apprehension marked by

heightened physiological arousal (Levitt, 1980). Currently, researchers believe anxiety to

be a multidimensional property (Martens et al., 1990), and more complex than this

definition implies. Much of the current research is based primarily on Spielberger's

(1966) state-trait theory of anxiety. Spielberger defined state anxiety as an immediate

emotional state that is characterized by acute feelings of apprehension, fear, and tension









and is accompanied by physiological arousal. Trait anxiety is a feature of personality and

is the predisposition to perceive non-dangerous situations as threatening and to react with

inappropriate or disproportionate state anxiety.

Martens et al. (1990) proposed that state anxiety is comprised of cognitive anxiety

(cognitive-worry) and somatic anxiety (emotional-arousal). Cognitive anxiety "is the

mental component of anxiety and is caused by negative expectations about success or by

negative self-evaluations" (Martens et al., 1990, p. 6) whereas somatic anxiety "refers to

the physiological and effective elements of the anxiety experience that develop directly

from autonomic arousal" (Martens et al., 1990, p. 6). In other words, somatic anxiety is

the perception of physiological arousal. Originally, Martens et al. (1990) suggested that

cognitive and somatic anxiety affects performance negatively, but this notion has been

challenged, and will be discussed later in this chapter.

Stress

Anxiety is often confused is stress. From a physiological standpoint, stress is a

neutral physiological response to some sort of stressor. A stressor might include physical

exercise or threat to bodily harm (Selye, 1975). This is a slight variation than that which

emanates from sport psychology perspectives where stress is defined in terms of a

process rather than a state or response as is found in exercise physiology, for example.

McGrath (1970) and Spielberger (1972) originally described the stress process, which

was later modified by Martens et al., (1990). The stress process is explained in terms of

an object demand, a perceived threat, and a state anxiety reaction. The object demand,

which is considered the stimulus, represents a situation or environment that might elicit

perceptions of threat or danger. The perceived threat is a subjective evaluation of the

object demand and the individual's response capability to that demand, and is followed









by a state anxiety reaction or response. Stress occurs if the perceived demand exceeds

the individual's ability to respond.

In summary, arousal represents alertness and physiological activation whereas

anxiety is a multidimensional property with both state and trait properties. State anxiety

has both cognitive and somatic components, which affect performance whereas trait

anxiety, a feature of personality, predisposes an individual to react with an inappropriate

state anxiety response. Finally, stress is defined a process which invokes a state anxiety

reaction. A common link typically exists between these concepts in that when an

individual is stressed, state anxiety is invoked, and arousal occurs.

Anxiety and Performance

The relationship between anxiety and performance is most readily observed by

examining pre-competitive anxiety. Researchers have attempted to study the temporal

changes in cognitive anxiety, somatic anxiety, and physiological arousal as the time to

the competitive event approaches. This can be accomplished through observation of

behavior indicators, electrophysiological measurement of arousal, or paper-and pencil

evaluative tests. The more common paper and pencil tests are the State-Trait Anxiety

Inventory (STAI) (Spielberger, 1983), Competitive State Anxiety Inventory-2 (CSAI-2)

(Martens et al., 1990) and the Sport Anxiety Scale (SAS) (Smith, Smoll, & Schultz,

1990). While results have varied, the primary findings are that cognitive anxiety starts

relatively high and remains high as the competitive event approaches (e.g. Jones & Cale,

1989; Jones, Swain, & Cale, 1991). Conversely, somatic anxiety remains relatively low

until approximately 24 hours before competition, and then increases rapidly as the event

approaches (e.g. Martens et al., 1990; Swain & Jones, 1992). During performance,









cognitive anxiety is expected to change continuously and is dependent on the current

situation variables, where as somatic anxiety is hypothesized to dissipate once

performance begins (Martens et al., 1990). Overall the research on the direction of pre-

competitive anxiety is fairly compelling. However because of the difficulty of measuring

real-time anxiety, there are no known studies that address competitive anxiety during the

event. Researchers and anxiety theorists are generally left with mere conjecture

concerning anxiety fluctuations during the course of the event.

Facilitating and Debilitating Effects of Anxiety

Even though the facilitating effects of anxiety are supported in anxiety-

performance literature, there are many researchers (e.g. Landers, 1994) who still view

anxiety as negative and debilitative to performance. In the late 1980's and early 1990's,

Hardy, Jones, and colleagues proposed that anxiety effects should be distinguished

between debilitative and facilitative states. This proposal is based on a large body of

literature from other areas of psychology, which suggests that anxiety should be viewed

as a broad continuum of states ranging from virtual immobilization to exhilaration

(Sarason, 1978). Also in test anxiety literature, the debilitating and facilitating dimension

of anxiety has been well known since the early 1960s (Jones, 1996). Yet, it was not until

the late 1980s and early 1990s that the notion of debilitative and facilitative anxiety states

became prominent in the sport psychology literature.

Support for the debilitative and facilitative distinction has been provided in

several recent empirical investigations and results are largely based on self-report data

gathered from administration of the CSAI-2. For example, Jones, Swain, and Hardy

(1993) used a sample of female gymnasts to examine the relationship between direction









and intensity of competitive state anxiety and performance on the balance beam. While

results showed no difference between good and poor performing groups with respect to

levels of cognitive and somatic intensity or direction anxiety, the good performance

groups reported their cognitive anxiety as being more facilitative to performance than did

poor performance groups.

Jones, Hanton, and Swain (1994), in a sample of elite and non-elite swimmers,

examined the importance of performance level as an individual difference variable. They

found, once again, no difference in the intensity of cognitive and somatic anxiety, but

found that elite swimmers interpreted both somatic and cognitive anxiety as being more

facilitative to performance when compared to non-elite swimmers.

Finally, Swain and Jones (1992) provided longitudinal evidence for the

facilitative and debilitative distinction with participants comprising a university

basketball team. Their results demonstrated that individual players could have very

different emotional states even with the same cognitive and somatic anxiety scores.

Individuals also reported having the same cognitive and somatic anxiety scores before

two different games, while having a positive experience facilitativee) in one and a

negative experience debilitativee) in the other.

With this strong empirical support, the distinction between debilitative and

facilitative anxiety represents a worthwhile pursuit for anxiety researchers that will lead

to further understanding of the competitive anxiety-performance relationship. However,

since the conclusions are based almost exclusively on the CSAI-2, they do not address

the mechanisms whereby anxiety can have a positive or negative consequence on

performance. Furthermore, continued research in identifying cortical activity and heart









rate correlates of anxiety variability is necessary to fully address the mechanisms that

influence the debilitative and facilitative distinction. From a theoretical standpoint, there

are only two theories that consider anxiety as having both positive and negative

consequences: the cusp catastrophe model of anxiety and the processing efficiency

theory. The following section contains a discussion and a critique of the current theories

of anxiety, with specific emphasis on the catastrophe theory and the PET.

Theories of Anxiety

The preceding section provided a basic introduction to the terms and definitions

that help distinguish anxiety, stress, and arousal. For a more complete understanding of

anxiety, it is important to examine both traditional and current anxiety theories. Jones

(1995), in a review of anxiety research, divided the popular theories into three general

approaches: (1) arousal-based, (2) general anxiety-based, and (3) multidimensional

anxiety-based. The following section is organized using the same distinctions set forth

by Jones.

Arousal-Based Approaches

Arousal-based approaches have the longest legacy in sport psychology research.

Drive theory, originally proposed by Hull (1943) and modified by Spence and Spence

(1966), and the inverted U-hypothesis, include arousal-based explanations to describe the

anxiety-performance relationship.

Drive Theory. Drive Theory (Hull, 1943; Spence & Spence, 1956) explains the

relationship between learning and arousal as well as performance and arousal. The

effects of arousal on a beginner may be different than on skilled performers. Drive

theory is represented by the following formula: Performance = Arousal X Habit

Strength. Drive theory is a complex stimulus-response theory of motivation and learning.









Increased arousal (drive) will elicit the dominant response. The response associated with

the strongest reaction potential is the dominant response. Early in learning of complex

tasks, the dominant response is the incorrect response because the potential for error is

greater. Late in complex task learning or for simple tasks, the dominant response is the

correct response.

The theory was extremely difficult to test and the examinations of the theory often

yielded conflicting results (Martens, 1971, 1973). The tests were also believed to

simplistic to explain sport performance (Fisher, 1976). Drive theory was replaced by the

inverted-U hypothesis as the favored conceptual framework of sport psychology

researchers.

Inverted U-hypothesis. Even though the inverted-U hypothesis is a classic model

to describe the arousal/performance relationship, it is still championed by many

researchers today. Critics of the inverted-U hypothesis point out that it is far too

simplistic to account for the complex relationship between arousal and performance

(Jones & Hardy, 1989). Although, some researchers, such as Landers (1998), still believe

that the inverted-U theory is the appropriate conceptual framework to describe the

influence of arousal on performance, and site Occum's razor (that is, not to involve more

ideas in a hypothesis or a theory than is necessary) as the reason they reached this

conclusion.

Simply stated, the relationship between performance and arousal is quadratic as

opposed to linear and takes the form of an inverted-U. The Inverted-U theory is founded

upon the classic work of Yerkes and Dodson (1908) using dancing mice as subjects. The

Yerkes-Dodson law states that for every behavior there is an optimal level of arousal.









This optimal level of arousal typically is of moderate intensity in order to produce

maximum performance. Furthermore, Yerkes and Dodson highlighted the role of task

difficulty. They suggested that an easily acquired habit, one that does not require

difficult sense discrimination or complex associations, may readily be performed under

strong stimulation (for example, a bench press in weight lifting). However, a difficult

habit may be acquired readily only under relatively weak stimulation (for example,

putting in golf). Thus, the optimal level of arousal varies as a function of the complexity

of the task and the skill of the athlete. Highly skilled athletes and athletes performing

simple tasks need a moderately high level of arousal for maximum performance. Less

skilled athletes and athletes performing a complex task require a relatively low level of

arousal for maximum performance.

Support for an inverted-U hypothesis is well documented in the literature for such

tasks as response time (Lansing, Schwartz, & Lindsley 1956), auditory tracking (Stennet,

1957), and hand steadiness (Martens & Landers, 1970). However, more recent evidence

points to a lack of clear empirical support for the hypothesis (Hockey, Coles, & Gaillard,

1986; Neiss, 1988). One of the difficulties in testing the inverted-U hypothesis with

humans is the inability to precisely measure arousal (Landers, 1994). For this reason, it is

difficult to refute the theory. For example, if researchers fail to demonstrate that

heightened arousal caused a decrement in performance, it could be argued that for the

particular task used, arousal was not high enough (Eysenck, 1984; Landers, 1980). If

arousal had been elevated sufficiently, higher performance would have declined.

Furthermore, support for the inverted-U hypothesis is only in the form of performance

outcome, which is not necessarily a direct measure of the complexity of the relationship









between arousal and performance (Eysenck, 1984; Eysenck & Calvo, 1992). Finally, one

of the largest criticisms of the inverted-U stems from the assumption that arousal is a

unidemensional property. Considering arousal a unidimensional property ignores the role

of cognitive, physiological, and behavioral components that affect the arousal-

performance relationship (Jones, 1995). In essence, the inverted-U is a simplistic concept

used to define the complex nature of anxiety and arousal.

General Anxiety-Based Approaches

State-trait anxiety theory. One of the more common general anxiety-based

approaches is Spielberger's (1966) state-trait anxiety theory, as measured by the State-

Trait Anxiety Inventory (STAI) (Spielberger, Gorsuch, & Lushene, 1970). The use of the

state-trait anxiety theory in sport research has resulted in the same type of conclusions as

found for the inverted-U hypothesis. Generally, both high and low levels of anxiety will

interfere with performance (Spielberger, 1989). The state-trait approach considers

unidimensional factors and suffers from the same criticisms as the inverted-U hypothesis.

Zone of optimal functioning. More recently, Hanin developed a general anxiety-

based approach used to describe the anxiety-performance relationship. Hanin's (1980,

1986, 1989, 2000) Zone of Optimal Functioning (ZOF) postulates that the level of

optimal state anxiety best for one athlete may be very different from the next athlete.

According to Hanin, if an athlete's optimal precompetitive state anxiety level can be

determined, it should be possible to help an athlete achieve that ideal level through

arousal control techniques. Using an adaptation of the STAI, an athlete's zone of optimal

functioning can be determined through repeated observation of an individuals'

precompetitive state anxiety levels, and an individuals' performance levels.









Support for this theory has been provided by Hanin and other researchers such as

Raglin et al. (1990) and Prapavessis and Grove (1991), where precise predictions about

precompetitive state anxiety and performance have been made. The theory also has a

level of intuitive appeal with sport psychologists and athletes because it represents an

attempt to locate the factors that influence peak performance on an individual basis.

However, the ZOF is based on a unidimensional approach to anxiety rather than a

multidimensional approach, and the ZOF cannot be used to explain underlying

mechanisms that influence performance (Jones, 1995). Also, the ZOF employs the STAI

as the primary measuring instrument, which is not a sport-specific measure.

Multidimensional Anxiety-Based Approaches

In the previous two sections, a common criticism among all of the models

presented is the lack of multidimensionality used to explain the anxiety-performance

relationship. Many current theorists believe that the complex nature of anxiety requires it

to be explained as a multidimensional property (e.g., Hardy & Parfitt, 1991; Jones, 1995;

Martens et al., 1990). With this in mind, the following three theories are presented: the

multidimensional anxiety theory, the cusp catastrophe model of anxiety and performance,

and the processing efficiency theory, with the major emphasis given to the processing

efficiency theory.

Multidimensional Anxiety Theory. Martens, Vealey, and Burton (1990) proposed

the multidimensional anxiety theory which predicts that cognitive anxiety has a negative

linear relationship and somatic anxiety has an inverted-U relationship (see Figure 2.1)

(Burton, 1988). Only Burton (1988) found support for both predictions; however, other

researchers, using regression analysis, have been able to test and offer partial support for









the dual model (e.g., Gould, Petlichkoff, Simons, & Vevera, 1987; Gould, Petlichkoff, &

Weinberg, 1984; Martens, Burton, & Vealey, 1990).


Hi Somatic Anxiety












Low


Low Hi

Anxiety

Figure 2.1 Multidimensional Anxiety Theory.

These results are based on data obtained from the Competitive State Anxiety

Inventory-2 (CSAI-2) (Martens, Burton, Vealey, Bump, & Smith, 1982, 1990), and only

represent precompetitive anxiety (Hardy et al., 1996). The CSAI-2 was originally

designed to measure cognitive and somatic anxiety, but during the validation process

self-confidence emerged as a third dimension. The typical finding is that self-confidence

is orthogonal to cognitive anxiety during times of stress, and that performers are able to

experience both self-confidence and cognitive anxiety simultaneously (Edwards &

Hardy, 1996; Hardy & Whitehead, 1984).

The multidimensional anxiety model is considered an impressive starting point for

understanding the anxiety-performance relationship, but it is not without criticism









(Hardy, 1996), and results have often been equivocal. There are four major criticisms of

the multidimensional anxiety model. First, the multidimensional anxiety model does not

predict an interaction between cognitive anxiety and somatic anxiety. Cognitive anxiety,

somatic anxiety, and self-confidence are considered additive rather than interactive.

Secondly, high cognitive anxiety is viewed as detrimental to performance, but recent

evidence (Edwards & Hardy, 1996; Jones, Hanton, & Swain, 1994; Jones & Swain, 1992;

Jones, Swain, & Hardy, 1993) has demonstrated that cognitive anxiety can have a

facilitative effect upon performance.

To illustrate the first two criticisms, Edwards and Hardy (1996) examined six

female netball teams and provided evidence for the interaction of physiological arousal

(somatic anxiety) and cognitive anxiety, as well as the facilitative and debilitative effects

of cognitive anxiety. The researchers used median splits on physiological arousal (heart

rate) and cognitive anxiety (from the CSAI-2) to divide the participant pool into four

groups: 1) high cognitive anxiety, high physiological arousal; 2) high cognitive anxiety,

low physiological arousal; 3) low cognitive anxiety, high physiological arousal; and 4)

low cognitive anxiety, low physiological arousal. After completing a two-factor analysis

of variance (cognitive anxiety X physiological arousal), Edwards and Hardy found a

significant interaction that matched their hypothesis. High cognitive anxiety and low

physiological arousal, and low cognitive anxiety and high physiological arousal both had

higher performance. Conversely, low cognitive anxiety and low physiological arousal

had a debilitating effect on performance, and low physiological arousal and high

cognitive anxiety had moderate effects on performance. Overall, their results supported

the hypotheses that high cognitive anxiety may either be facilitative or debilitative to









performance when accounting for the level of physiological arousal, and that there is an

interaction between physiological arousal and cognitive anxiety.

Third, the multidimensional model follows the basic assumptions of the inverted-

U hypothesis in that small incremental increases in arousal result in small incremental or

decreases in performance and moderate arousal results in optimal performance. Anxiety

theorists (Hardy, 1996; Jones, 1995) have challenged this notion and further questioned

the assumption that if an athlete's anxiety level is too high, restoring it to median level

will improve performance. Finally, the multidimensional anxiety model is based on and

supported by the CSAI-2, which only measures precompetitive anxiety. There is no

attempt to explain underlying mechanisms or the precise effect that cognitive and somatic

anxiety have on performance effectiveness.

Cusp catastrophe model of anxiety. The cusp catastrophe model of anxiety and

performance (Hardy & Fazey, 1988) was originally formulated as a challenge to the

inverted-U hypothesis and represents an attempt to explain some of the criticism aimed at

the inverted-U hypothesis and at the multidimensional anxiety model. In the

multidimensional anxiety model, it is proposed that somatic anxiety and cognitive anxiety

are independent with different antecedents. However, in the cusp catastrophe model, it is

proposed that cognitive anxiety and physiological arousal (rather than somatic anxiety)

are not independent but interact with each other to influence performance. Finally, the

cusp catastrophe model also predicts the potential facilitative and debilitative effects of

anxiety and does not consider anxiety to always be a detriment to performance.

The cusp catastrophe model (Thom, 1975) is a mathematical model used to

describe a wide range of discontinuous and divergent phenomena (Hardy, 1996), and was









originally popularized by Zeeman (1976) for use in the behavioral sciences. It is used to

describe, mathematically, discontinuities that occur in the physical world that are

typically continuous (Hardy, 1996). The cusp catastrophe model of anxiety and

performance (see Figure 2.2) was originally proposed and applied to a sport setting by

Hardy and Fazey (1987).

The model is three-dimensional, which "demonstrates how one dependent

variable can demonstrate both continuous and discontinuous changes as a result of

continuous changes in two other independent variables" (Hardy, 1996, p. 142). The basic

premise is that if anxiety increases beyond the optimal level, instead of a gradual drop off

in performance (as in the inverted-U and multidimensional anxiety model), there is a

significant and dramatic reduction in performance. The three interacting dimensions of

the model are physiological arousal, cognitive anxiety, and performance.

First, Hardy uses physiological arousal (measured by heart rate) instead of

somatic anxiety to represent the asymmetry factor, or the normal factor (Hardy, 1996).

The asymmetry factor determines how close the dependent variable (performance) is to

the drop-off point. Physiological arousal is an objective response rather than a subjective

response as in the case of somatic anxiety. According to Hardy (1996), physiological

arousal influences performance by potentially reducing the available cognitive and

physiological resources to performers, and through positive or negative interpretations of

their physiological symptoms, whereas somatic anxiety has been hypothesized to only

influence performance "if the magnitude of the somatic response is so great that the

performer becomes preoccupied with his or her perceived physiological symptoms"

(p.142).









The second issue is how cognitive anxiety interacts with physiological arousal

and performance. Hardy (1996) defines cognitive anxiety as the splitting factor, that

which controls the effects of the asymmetry factor on the dependent variable. In this

case, the level of cognitive anxiety determines the degree to which physiological arousal

effects performance. Basically, physiological arousal is always present, and will only

affect performance if levels of cognitive anxiety are also considered.



Performance





Cognitive X1
Anxiety X2

SPhysiological Arousal


Figure 2.2: Cusp Catastrophe Model (Hardy, 1996).


The relationship between the three dimensions can be summarized as follows: (1)

during periods of low cognitive anxiety, there will be a gentle inverted-U relationship

between performance and physiological arousal; (2) during periods of high cognitive

anxiety, performance will improve as long as physiological arousal does not reach a

critical threshold, after which any further increase in physiological arousal will result in a

dramatic and abrupt decrease in performance; (3) a negative correlation is predicted

between cognitive anxiety and performance when physiological arousal is high; (4)

conversely, a positive correlation is predicted when physiological arousal is low. It

should be noted that, the performance curves which represent the upper and lower









performance surfaces are defined as opposing curves: the upper one representing

performance during increasing physiological arousal (Xl); and the lower one

representing performance during decreasing physiological arousal (X2). Thus, depending

on the current level of physiological arousal, high cognitive anxiety can either have a

positive or negative effect on performance until such a point in which increasing

physiological arousal causes a dramatic and sudden decline in performance. Once the

deterioration has occurred, only a considerable reduction in physiological arousal beyond

the drop off point can return performance to its previous level. This situation is referred

to as "hysteresis" and the result is a "bifurcation set" in which the same level of

physiological arousal is associated with two different levels of performance (Hardy,

1996).

While the cusp catastrophe model has been the object of little empirical

investigation, its predictions have received some empirical support in studies on

basketball players (Hardy & Parfitt, 1991), softball players (Krane, Joyce, & Rafeld,

1994), and bowlers (Hardy, Parfitt, & Pates, 1994). For example, Hardy et al. (1994)

attempted to manufacture anxiety in a controlled setting in which participants were

experienced crown green bowlers (British lawn bowling), and they were placed into two

groups: ego-threatening instructional set or a neutral set condition. The participants in

the ego-threatening instructional set group were instructed that their scores would be

compared to elite crown green bowlers, and they would need to perform well in

comparison to these bowlers. The manipulation of physiological arousal was through

physical exercise (shuttle runs). Comparisons were made between low cognitive anxiety

and high cognitive anxiety, and heart rate direction (increasing or decreasing). A three-









way repeated measures analysis of variance was performed, and it was concluded that the

high cognitive anxiety with heart rate increasing condition did significantly better than

the performance in other conditions which represented support for the model.

The cusp catastrophe model is considered an innovative approach (Jones, 1995),

but is not without criticism. First, Gill (1994) criticized the model's complexity, over-

sophistication, and difficulty in testing. Furthermore, many predictions of the model are

made during competition and it is difficult to prove or disprove those predictions because

much of the evidence to support the model is based on data from the CSAI-2. Gill also

criticized an apparent lack of practical application. Second, the model does not offer any

suggestions about the underlying mechanisms that influence behavior and/or

performance. Its predictions are based on global performance measures with potential

minute distinctions that may not be sensitive enough to allow for the detection of anxiety

effects. Thus, it becomes a model, which is not only difficult to test, but difficult to

prove or disprove. However, recent developments in identifying positive and negative

emotions through the use of psychophysiological measures offers optimism for

proponents of the cusp catastrophe model.

Processing efficiency theory. The Processing Efficiency Theory (PET) has

primarily been applied to test anxiety. However, the PET is sufficiently broad that it

could have wide application in sport psychology and motor learning. The theory is

designed to account for individual differences in task performance in stressful conditions,

and the relationship of state anxiety and performance in these situations. PET also

attempts to account for the debilitative and facilitative effects of anxiety.









The predictions of the PET are based partially on the role of state anxiety and the

assumption that the human brain has only a limited capacity for information processing.

State anxiety is operationalized to result from the interaction of trait anxiety and the

situational threat or the environmental demand. Worry (cognitive anxiety) in the form of

self-preoccupation, concern over evaluation, and concern over level of performance is

operationalized to be the cognitive component of state anxiety. Cognitive anxiety is

hypothesized to occupy some of the capacity of the central executive of the working

memory system (Baddeley, 1986) leaving fewer resources for task related processing.

The working memory system, which represents an attentional control system,

consists of three capacity limited components: a modality-free central executive, an

articulatory loop, and a visuo-spatial sketchpad. The central executive is responsible for

the allocation of resources and coordinates short-term memory systems (the visuo-spatial

sketchpad, and the articulatory loop). It is therefore primarily concerned with the control

of action and the integration of information (Baddeley, 1993). The articulatory loop is

capable of holding and manipulating speech-based information while the visuo-spatial

sketch pad deals with visual and spatial information. As stated previously, cognitive

anxiety impacts the central executive, but it also impacts to a lesser extent the articulatory

loop. The articulatory loop is responsible for verbal information, and cognitive anxiety is

thought to be non-vocalized, verbal information and therefore is processed in the

articulatory loop. The impact on these systems will generally negatively affect

performance especially in situations of high anxiety. As cognitive anxiety and task

difficulty increases, an increase in processing resources and storage resources is required

in the central executive. Therefore, when an individual experiences high levels of









anxiety, cognitive resources are impacted, which decreases the amount of processing

resources available for the task, thus reducing task performance.

According to Eysenck and Calvo (1992), the above explanation does not account

for any increased motivation or extra effort applied to the task. Hockey (1986) proposed

a control unit that works directly with the central executive in the working memory

system, and is hypothesized to effect the capacity changes induced by anxiety. The

control unit alters processing resources to allocate more processing toward strategies to

improve performance, and acts as mediator of anxiety on processing and on performance

(Eysenck & Calvo, 1992). The negative consequences of poor performance and the

knowledge of these consequences can promote an increase in effort by allocating

additional resources to the task. The theory follows that if a situation is aversive due to

anxiety it will be avoided, but if the situation is not avoided then the allocation of

additional resources (Eysenck & Calvo, 1992) can lead to increased or maintained

performance. Eysenck and Calvo suggested that high trait anxious performers will

devote more processing resources to negative thoughts, worry, and cognitive anxiety.

Thus, the control unit reacts to the resource increase and applies more effort to the task.

Through more effortful processing for the task, high anxious individuals transfer control

from lower automatic subsystems to higher controlled subsystems (Hardy, Mullen, &

Jones, 1996).

Because the control unit allocates extra processing resources to the task, it is

therefore important not only to measure performance effectiveness but also processing

efficiency. Performance effectiveness is easily measured and is considered the outcome

of the performance (e.g., task score). Processing efficiency is described as the division of









performance effectiveness by effort, while effort is the amount of processing resources

dedicated to the task. Anxiety will affect processing efficiency more than performance

effectiveness. Based on this conclusion, Eysenck and Calvo offer two predictions: (1)

cognitive anxiety requires processing resources and reduces available resources for the

task; (2) the presence of anxiety will cause an increase in effort for the task designed to

maintain or to improve performance. In other words, there will be more attention applied

to the task than is otherwise needed. Furthermore, Eysenck and Calvo point out that

anxiety will differentially effect processing efficiency and performance effectiveness, and

it is important to measure these separately.

To summarize, Eysenck and Calvo described four consequences, which cause an

increased role by the control unit. First, cognitive anxiety, which is influenced by current

performance effectiveness, can cause an increase in motivation designed to improve

performance and decrease cognitive anxiety while increasing effort. As stated previously,

Jones (1995) and Jones, Hanton, and Swain (1994) have demonstrated that an

individual's interpretation of anxiety can lead to either improved performance or

decreased performance. If an individual interprets anxiety as positive (i.e., motivation),

this will lead to improved performance (facilitating), and conversely, a negative

interpretation will lead to a performance decrease (debilitating). These findings

demonstrate that anxiety can have a positive effect and lead to facilitation of

performance. Furthermore, from a PET perspective individuals might report a positive

disposition towards anxiety, but in essence, anxiety provides an increase in motivation,

which is a slight theoretical deviation from Jones et al.'s conclusions.









Second, high-anxious individuals will allow anxiety and the threat of the

impending performance to consume a greater amount of processing resources than low-

anxious individuals. Third, not only do high-anxious individuals increase processing

resources for anxiety, they are also quicker in detecting situations as threatening, and will

selectively allocate more attention to those events. This is due to the notion that high-

trait anxious individuals have a higher probability of perceiving events to be threatening

although these events might not be. Finally, high-trait anxious individuals may set

unrealistically high goals for performance outcomes, which will increase probability of a

performance-expectation mismatch.

Overall, research in test anxiety literature involving the PET has demonstrated the

role of the working memory system in anxiety, and has tested the predictions of

processing efficiency and performance effectiveness (e.g., Elliman, Green, Rogers, &

Finch, 1997). In sport psychology, the PET has only received cursory interest (e.g.,

Parfitt & Hardy, 1993) with no known study directly testing the predictions of the PET.

Thus, a lengthy discussion of the many studies that test different aspects of the PET is not

warranted. However, the following studies offer support for the processing efficiency

and performance effectiveness distinctions, and appear to be the most relevant to this

investigation. Following the exemplary studies presented here is a discussion of the

predictions of the PET, which includes further evidence supporting the theory. Because

the goal of this investigation is to test the processing efficiency theory using response

time, visual search, and electroencephalogram (EEG) measures, predictions of the theory

are discussed in light of these dependent measures.









The typical approach in testing the processing efficiency-performance

effectiveness distinction is through the use of a dual task paradigm where anxiety is not

manipulated. Comparisons are made between high and low trait anxiety individuals.

Following a description of each study are some general comments and criticisms about

this line of research

Dibartolo, Brown, and Barlow (1997) demonstrated support for the effort and

effectiveness distinction. In this study, comparisons were made between individuals with

general anxiety disorder (GAD) and normal non-anxious individuals using a trigram

reaction time task varied by task importance. The GAD and control groups significantly

differed on the baseline trials, which is expected in this type of task. However, as task

importance increased, the GAD group exhibited an increase in worry and anxiety as well

as lower performance expectations prior to performing the task, but their performance

increased relative to baseline, and was not significantly different than controls. Due to an

increase in anxiety, the GAD group applied more processing resources to the task and

was able to subsequently improve their performance.

In addition, work by Elliman, Green, Rogers, and Finch (1997) and by Ikeda,

Iwanga, and Seiwa (1996) supported the predictions of the PET by demonstrating that

high anxious individuals will have increases in processing times, but no significant

changes in accuracy. In Elliman et al. (1997), participants were divided by three anxiety

groups (high, medium, and low anxiety) using the Hospital Anxiety and Depression Scale

(Snaith & Zigmond, 1994). A sustained attention task, a reaction time task, and a motor

speed task were used as dependent measures. Their results indicated that individuals

differed only on reaction time for the sustained attention task, which was considered the









measure of efficiency. However, the three groups did not differ on the accuracy of the

sustain attention task nor on the motor speed task or the reaction time task.

Further evidence by Ikeda et al. (1996) included a verbal memory task in which

high test-anxious participants required longer to process information than low test-

anxious participants, while both groups achieved the same level of accuracy on the task.

This is consistent with other investigations in that for the most part, support for the PET

has been in the form of differences in reaction time without changes in accuracy (e.g.

Elliman et al., 1997).

While the results of the previous studies are compelling, they only offer

elementary support for the theory. The potential physiological or underlying cognitive

changes that may occur due to anxiety have not been addressed. Support for the theory

involves only speed and accuracy changes along with increases in task difficulty. By

limiting assessment to these variables, prior researchers have ignored potential factors

such as heart rate, or electroencephalogram (EEG), which might be more direct indicators

of the efficiency and effectiveness distinction. Thus, there is a need to provide a more

comprehensive account of processing efficiency. Finally, the majority of research

involving the PET (Eysenck, 1996) has included tasks with a substantial verbal

component, with little consideration for how anxiety influences performance on motor

tasks. The following three sections offer an account of how to test the theory using

reaction time, visual search, and EEG as dependent measures.

Reaction Time and Orienting of Attention

In an effort to offer a better understanding of orienting of attention and its

relationship to the processing efficiency theory, a short summary of attention

characteristics will be followed by a discussion on orienting of attention. Attention can









be conceived as consciousness (Cohen & Schooler, 1997), effort or arousal (Beatty &

Wagoner, 1978), or as capacity or resource (Broadbent, 1958; Kahneman, 1973). Since

one of the assumptions of the PET is that attention is a limited resource, the following

summary emanates from that perspective.

Since at least the early 1950s, one of the major assumptions in attention related

research is that attentional capacity is limited in some way. This limited capacity can be

most readily seen in the dual task paradigm. In this paradigm, subjects are given a

primary and a secondary task, which both require some allocation of attention. The logic

follows that as the attentional requirements of the primary task increase, attentional

allocation to the secondary task decreases resulting in decreased performance of the

secondary task. Several models have been proposed, but there are still debates as to

which theory best describes attention. Despite widespread debate, these theories can be

placed into two basic categories: filter theories and resource model theories.

Broadbent (1958) originally proposed the first filter theory, which assumes that

one can allocate attention to different inputs or tasks, and possess the ability to filter

extraneous or unwanted information. According to this view, attention is thought of as a

single resource and some point in the processing of the incoming information irrelevant

stimuli or information are eliminated before it reaches memory. There are now several

filter theories (Broadbent, 1958; Deutsch & Deutsch, 1963; Norman, 1968), which differ

on the point at which irrelevant information is eliminated.

Resource model theories contrast to filter theories in that attention is viewed as a

limited-capacity resource (Kahneman, 1973). In the case of Kahneman's resource model

theory, attention is hypothesized to be as a single 'pool' that allows execution of several









tasks as long as the capacity of the 'pool' is not reached. Once the capacity of the

attentional resource is exceeded, task performance begins to deteriorate. The available

capacity will fluctuate as the level of arousal and the demands of the task or tasks change.

In addition to Kahneman's model, other theorists have proposed multiple resource

models, in which there is no single attentional resource, but attention is divided into

several distinct subsystems or pools of resources (Navon & Gopher, 1979). Multiple-

resources models were developed to explain the lack of performance decrements found in

studies in which multiple tasks differed in modality and the response for each task. For

example, Wickens (1976) had participants perform a manual-tracking task while

simultaneously responding to an auditory tone or maintaining constant pressure on a

stick. Curiously, responding to the auditory tone produce less detriment to the primary

task. It was concluded that since the modalities were different, there was no interference

and attention was divided among tasks. Multiple-resource models differ on how the

pools are divided and the stages of processing in which attention is engaged (Navon &

Gopher, 1979; Wickens, 1984).

At present, debate remains concerning as to the exact nature of attention and

which theory best reflects what is occurring. However, the processing efficiency theory

is based partially on Kahneman's single resource model theory. That is, it is assumed

that capacity is limited and there is a single attentional 'pool' that can be impacted by

anxiety and arousal. As stated earlier, capacity limitations are demonstrated through the

dual task paradigm as well as the processing efficiency and performance effectiveness

distinction. Thus, this study will employ a dual task paradigm. The following section









examines the primary task, which involves participant driving a racecar simulator, and

the reaction time task, as they relate to processing efficiency.

Driving and the PET

The use of a driving simulator is an ideal way to test many of the predictions of

the PET. Simply, driving is a well-practiced task, which requires few cognitive resources

in normal conditions. While the simulator to be used in this study is different from actual

driving and performance will be based on speed on a simulated racecourse, the

generalizations about driving and cognitive resources still apply. Most drivers have

developed a level of automaticity for driving, thus decreasing the resources needed for

the task and increasing the available resources for secondary tasks. Under normal

conditions, the average driver can, and typically does, take on many tasks other than

driving these might include eating, changing the radio station, talking with passengers,

and the like. However, during situations of increased cognitive load and task difficulty

such as increased traffic, bad weather, and fatigue, an individual will attempt to maintain

performance while avoiding driving error. Generally, individuals will be able to maintain

performance even under increased cognitive load and task difficulty as long as capacity

limits are not reached.

Once capacity limits have been reached and mental workload is maximized, the

potential for errors in driving performance dramatically increases. For example, Zeitlin

(1995) examined drivers who performed secondary auditory tasks on a variety of roads to

determine workload maximums and error production. Results indicated that increased

traffic, and adverse weather conditions resulted in a reduction of performance on the

secondary task, and an increase in subjective difficulty ratings. Furthermore, the adverse

driving conditions produced an increase in brake depressions. While not reported in this









study, it could be argued that an increase in traffic and a decrease in the automaticity of

the task caused the increase in brake depressions.

Miura (1990) examined participants' useful field of view and peripheral visual

performance in a real driving setting. The useful field of view is considered the

information-gathering area of the display. Using a reaction time response to peripheral

targets, Miura was able to demonstrate a narrowing in the useful field of view, longer

reaction times, and an increase in saccadic eye movement as the task demands increased.

Miura concluded that the results indicated deeper processing at each fixation point which

caused the useful field of view to narrow, and a decrease in peripheral performance as

task demands increased. While Miura did not manipulate anxiety, it is possible that

anxiety or stress from increased task demands effected the results. Specifically, anxiety

or stress consumed cognitive space, and reduced spare processing space for peripheral

detection and the reaction time task, thus required the participants to allocate more

resources to driving. Furthermore, the decrease in spare processing space could have also

limited the useful field of view and contributed to a narrowing of attention (Janelle et al.,

1999).

In summary, driving is generally a well-practiced skill, which during normal

driving conditions does not require much processing capacity. However, in moments of

high stress and when anxiety is present (as will be the case in the current study), the

potential for error will increase, and thus participants will need to apply more cognitive

resources as allocated by the control unit in the central executive to the primary task to

maintain driving performance. The increase in effort in the primary task will cause a









decrease in performance in the secondary task. The secondary task, and its role in

orienting of attention, is discussed in the following section.

The Secondary Task and Orienting of Attention

Reaction time assessment of processing efficiency and the secondary task is

typically conducted with a probe technique. Eysenck and Calvo offer the prediction that

"anxiety will reduce spare processing capacity during the performance of a central task"

(p. 420) as assessed by a probe technique. The probe technique is a secondary task that

requires participants to respond as rapidly as possible to an occasional stimulus while

performing a central task. The probe technique requires allocation of processing

resources in which changes in reaction time offer a direct measure of spare processing

capacity. Eysenck (1989), in a letter-transformation task, demonstrated that while

accuracy on the primary task did not change for high and low anxious performers,

reaction time on the probe task was greater for high anxious individuals. Since the

processing efficiency theory assumes a fixed attentional capacity, the probe task reveals

changes in this capacity. Processing resources available to each task and size of the

attentional focus will modulate these changes.

In order to examine attentional focus and processing efficiency, a cost-benefit task

(Posner, 1980) is often employed as the probe task. The cost-benefit paradigm has been a

common method for examining orienting of visual attention (Nougier & Rossi, 1999),

attentional focus, and processing efficiency (Castiello & Umilta, 1990). Posner et al.

(1978) demonstrated the costs and benefits of anticipation. In this investigation,

participants were required to fixate on the center of the screen and would then receive

one of three precues. Participants were then to respond as quickly as possible to one of

two targets. Precues included an arrow pointing either left or right, or a "plus" sign in the









middle of the screen. Arrows were considered directional precues and were either valid

or invalid. A valid precue directed the participant in the correct direction to the target

and an invalid precue incorrectly directed the participant to the wrong target area. Valid

and invalid cues occurred for two-thirds of the trials whereas the "plus" signs, which

were neutral precues, occurred for one-third of the trials. Arrow precues were correct

80% of the time and incorrect on 20% of the trials.

Results indicated that there was a 30 ms benefit of the valid trials when compared

to the neutral trials, and there was a 39 ms cost of the invalid trials when compared to the

neutral trials. Thus, orienting attention correctly increases processing efficiency. In other

words, the stimulus when given correct information is processed with a greater

efficiency, which affords an attentional benefit. The converse occurs when the

information is incorrect, which provides an attentional cost. The cost-benefit paradigm is

well researched and well supported (Nougier & Rossi, 1999), and is ideally suited for the

sport psychology and motor learning context in that it offers evidence for attentional

mechanisms that are needed and used in sport performance.

In sport performance, attentional cost and benefit depends greatly on how well the

performer anticipates based on the information given. Generally, attention is voluntarily

allocated to a certain location, which inhibits the processing of information presented in

an unattended location (Posner, 1980). Allocation or orienting of attention can have both

a positive and a negative effect on performance. If the allocation of attention brings

pertinent information to the task, it will help improve performance, which is especially

helpful in situations that require rapid decisions to be made and when the external context

is continually changing (Nougier & Rossi, 1999). Conversely, if attention allocation is









not focused toward useful information or if attention is divided from the primary task

through internal or external distraction, it will become detrimental to performance. The

detriment will be higher when the situation is relatively stable (for example a golf putt),

and any disturbance can cause a reduction in performance. Since voluntary allocation of

attention is not absolute, and it is influenced by internal and external distraction, there is

evidence that it can be automatically controlled (e.g., Jonides, 1981; Remington &

Johnston, 1992).

Originally, Jonides (1981) postulated that centrally located abrupt onset cues in

the cost-benefit design captured attention voluntarily whereas the peripherally located

abrupt onset cues captured attention in an involuntary, stimulus-driven manner. To test

this notion, Jonides (1981) conducted three experiments comparing central and peripheral

cues. Experiment 1 tested the attention capturing power of the peripheral cue with a dual

task paradigm. Participants engaged in a concurrent memory task while performing

visual search to central or peripheral cues. Peripheral cues produced attentional benefits

regardless the size of the memory load whereas response times to central cues were

affected by memory load. Thus, peripheral cues were relatively unaffected by increased

demands on the processing capacity indicating that these cues automatically captured

attention.

In Experiment 2, Jonides hypothesized that peripheral cues would be resistant to

suppression. That is, when given the instruction to ignore cues (central or peripheral)

participants would not be able to ignore peripheral cues due to the relative automaticity

associated with peripheral detection. When central and peripheral cues were randomly

presented, participants were able to ignore central cues, but not peripheral ones. Valid









peripheral cues represented the shortest reaction time and invalid peripheral cues

produced the longest reaction times. Curiously, the results in Experiment 1 and

Experiment 2 represented different magnitudes of costs and benefits, which was

dependent on the cue validity. In Experiment 1, cue validity was 70% whereas in

Experiment 2 cue validity was 12.5%. The changes in cue validity along with

corresponding changes in magnitude on both central and peripheral cues suggests that

participants do have some level of control over whether they attend to the peripheral cue,

but as evidenced in Experiment 2, participants do not have total control.

Experiment 3 tested whether expectancy of the cue presentation could influence

automaticity of the peripheral cues. Two groups were presented visual search trials in

which one group received 80% central cues and 20% peripheral cues while the other

group received 20% central cues and 80% peripheral cues. The results indicated that a

peripheral cue retained its effectiveness regardless of its probability of occurrence. The

general conclusion is that a peripheral cue is a "strongly automatic" process, and that it is

generally not susceptible to cognitive load and for the most part not subject to voluntary

control (Palmer & Jonides, 1988). However, the magnitude of the costs and benefits are

affected by the cue validity (Jonides, 1981). In addition, highly focused attention can

disrupt the effects of abrupt onset cues (Yantis & Jonides, 1990). Thus, if attentional

focus is modulated to some degree by anxiety, it seems plausible that anxiety will play a

role in determining the magnitude of the attentional costs and benefits, and could also

affect an individuals' attentional flexibility.

Attentional flexibility is conceived as the ability to move attention from one

spatial position to another, and to vary the amount of visual attention given to a particular









point in the scene (Castiello & Umilta, 1992). A large body of evidence exists that

demonstrates how attentional focus and flexibility are influenced by both endogenous and

exogenous factors (Nougier, Azemar, & Stein, 1992). Endogenous factors such as age

(Akhtar & Enns, 1989; Nougier et al., 1992) and exogenous factors such as task demands,

and attentional activation from sport practice, effect both magnitude of the attentional

costs and benefit as well as the attentional flexibility of the individual. LaBerge (1983),

and Eriksen and St. James (1986) both found that attentional focus varies by task

demand.

However, the role of processing efficiency is less clear. LaBerge (1983) used a

dual task situation to test spatial extent of focused attention. The primary task was to

identify the middle letter of five letters that formed a non-word or to identify the five

letters as a word. The secondary task was a probe task in which the participants were to

respond to a specific digit when it appeared in one of the five possible letter positions.

During the presentation of non-words, attention was narrowly focused on the middle

letter and reaction times to the probe were slower when the digit was presented in any of

the other four positions. When the attentional focus was wide to cover all five-letters, the

response to the probe was invariant across the five positions. LaBerge concluded that the

size of attentional focus could therefore vary due to task demands.

Eriksen and St. James (1986) using an eight-letter circular display tested the size

of attentional focus and processing efficiency. The number of precued locations and the

distance of distracters from the precues were manipulated. They revealed that if

attentional focus is increased to include all precued locations then the speed of processing

should decrease as the size of precued area increases (or attentional focus increases).









Eriksen and St. James found support for this hypothesis and for the notion that processing

efficiency is affected by the size of attentional focus, (i.e., the zoom lens).

Further empirical evidence for attentional focus was found by Castiello and

Umilta (1990) using a modified cost and benefits paradigm. Rather than having arrows

as cues, Castiello and Umilta used boxes that varied in size and they only used neutral

and valid trials. During valid trials one box would appear on the left or the right of the

center warning cross whereas during neutral trials both boxes would appear. As the size

of the box increased, reaction time also increased. Thus, Castiello and Umilta concluded

that processing efficiency decreases when the area of attentional focus increases. In light

of these three studies, the following conclusions can be made: (1) attentional focus varies

in accordance with task demands, (2) processing efficiency is inversely related to

attentional focus size, and (3) participants are able to voluntarily vary the size of their

attentional focus.

The above conclusions led sport psychology researchers to examine the

relationship of sport expertise and orienting of attention. With practice, open skilled elite

athletes seem to develop the ability to effectively reduce distractions from internal and

external sources, and also develop the ability to selectively attend to the most salient cues

in a scene. Nougier and Rossi (1999) point out that experts in open skill sports have

greater attentional flexibility for orienting their attention in visual space and are better

able to regulate their attentional resources depending on the specific task demands.

Several studies have demonstrated that attentional benefits and costs are smaller for elite

athletes when compared to novices in voluntary (Castiello & Umilta, 1992; Nougier,

Ripoll, & Stein, 1989; Nougier, Stein, & Bonnell, 1991) and automatic (Enns &









Richards, 1997; Nougier, Rossi, Alain, & Taddei, 1996; Pesce Anxeneder & Bosel, 1998)

orienting of attention. Nougier and Rossi (1999) hypothesized that elite athletes are

inclined to attend less to highly probable events and pay more attention to less probable

events. In effect, experts are more apt to expect the unexpected, while novices' focus is

more dependent on the highly probable events. Empirical support for these notions has

been shown with expert performers in sport skills such as fencing (Nougier, Stein, &

Azemar, 1990), boxing (Nougier, Ripoll, & Stein, 1989), volleyball, and tennis (Castiello

& Umilta, 1988; Nougier, Stein, & Bonnel). In summary, through practice, individuals

not only increase attentional flexibility, but also increase the overall processing space for

the task, and lessen the effects of anxiety on task performance (Eysenck & Calvo, 1992).

Anxiety, Visual Search and Visual Attention

Justification for the inclusion of visual search monitoring to offer evidence for the

processing efficiency theory is twofold. First, many cognitive tasks require selective

allocation of limited attentional resources (Folk & Hoyer, 1992). Visual search and

target identification requires shifts of attention and the distribution of resources across

display locations and environmental factors. As stated previously, anxiety is

hypothesized to affect the processing space available for the task. Thus, with limited

processing space available, the presence of anxiety may decrease the efficiency or the

speed with which attention is shifted to target locations. Second, changes in search rate,

fixation location, and fixation duration could provide evidence for increased effort and

decreased processing efficiency.

Visual search as detected by eye movement recording systems has become a

popular method for attaining information about cue usage (e.g., Bard & Fluery, 1976;

Bard, Fleur, Carriere, & Halle, 1980; Williams, Davids, Burwitz, & Williams, 1994).









The bulk of research involving sport and visual search has been conducted in the context

of the expert-novice paradigm. In many studies, there have been no differences found

between experts and novices (e.g., Abernethy & Russell, 1987) in fixation location, but

the general conclusions are that experts make better use of the information extracted.

While there are limitations in the use of eye tracking systems, it is still considered a

viable method for ascertaining cue usage and information pick-up (Williams et al., 1999).

Attention allocation is generally measured through covert shifts of attention as

indexed by the cost-benefit paradigm, described previously. However, in this study the

allocation of attention will be followed by a goal directed saccade, which necessitates a

shift from covert attention to an overt eye movement. Thus, the performer must first

detect information with peripheral vision and then bring in the information through foveal

vision. Not only does this require selective attention; it also requires a visual fixation.

One of the major debates in visual search research is the difference between a

visual fixation and selective attention (Williams, Davids, & Williams, 1999). It is

generally assumed that visual orientation is related to information extraction and visual

attention (Abernethy, 1988). However, it is possible to relocate attention without making

a saccade to a different location in a visual display (Williams et al., 1999). In other

words, participants will fixate on a specific location (which is commonly referred as

"looking") and can extract information from the same location or from a different

location (which is commonly referred to as "seeing") (Abernethy, 1988). While it is

impossible to execute a voluntary saccadic movement without a shift in attention, it is

possible to shift attention without making a saccadic eye movement (Shepherd et al.,

1986). Thus, currently in eye recording studies, it is not completely possible to know









where attention is directed by simply knowing the location of the eyes. However, there is

substantial evidence that demonstrates a close relationship between oculomotor and

attentional systems (Theeuwes, 1998). By requiring a goal-directed saccade in the cost

and benefit paradigm, it is possible to control not only the direction of attention, but also

the direction of subsequent saccade. However, the question arises of whether it is

possible to decouple spatial attention and saccadic eye movement.

Shepherd, Findley, and Hockey (1986) manipulated spatial attention by directing

participants to peripheral probe stimuli by varying the probability of each position, and

demonstrated that when the probe stimuli location and the location to which saccades

where made were similar, reaction times were shorter. Furthermore, Shepherd et al.

(1986) also provided evidence favoring the coupling of saccadic eye movement and

attentional enhancement. For example preparing to make a saccade to one location

produces an attentional enhancement to that same location.

Similarly, Hoffman and Subramaniam (1995) examined the relationship of

saccadic eye movements and covert orienting of visual spatial attention. Subjects were

required to attend to a particular location through directional cueing, and to make a

saccade to the same or a different location to detect a visual target. When the location of

the saccade and the location of attention were the same, target detection was highest.

Hoffman and Subramaniam proposed that when attention and the saccade are directed to

different locations, participants are faced with classic dual task interference.

Furthermore, Hoffman and Subramaniam's findings provide evidence that the allocation

of visual attention and saccadic eye movement are spatially coupled. It is important to

note that since attention precedes saccadic movement, fixation duration does not only









represent cognitive processing (Theeuwes, et al., 1998), but also preparatory time to the

next fixation location.

Kowler, Anderson, Dosher, and Blaser (1995) found further evidence that

attentional control and saccadic goals could not be separated using a letter identification

task. These findings offer evidence that voluntary-goal directed saccades and attention

cannot be split amongst locations, and that attention precedes the eyes to the saccade

target position (Theeuwes et al., 1998). Thus, it becomes difficult to decouple attention

and saccadic movement and evidence strongly suggests that visual attention and saccadic

eye movement are indeed obligatorily coupled (Schneider & Deubel, 1995). Therefore,

allocation of attention through spatial cueing results in more accurate and faster

processing of information in the space surrounding the cued region. It is expected that

the modified cost-benefit paradigm will produce an attentional enhancement and faster

saccadic movement when the locations are the same (Hoffman & Subramaniam, 1995),

and conversely when the locations are different should produce interference similar to a

dual task experiments.

Eysenck and Calvo (1992) offer the prediction that psychophysiological measures

can detect changes in efficiency more than effectiveness in the presence of anxiety.

Changes in fixation duration, fixation location, and increases in search rate could offer

evidence for increases in mental effort or an impairment of processing efficiency.

Janelle, Singer, and Williams, (1999) examined distraction and attentional narrowing in a

dual task paradigm using reaction time and driving performance. They found that

increases in anxiety led to an increase in saccadic and fixation activity, and increases in

distractibility to irrelevant peripheral cues. In addition, anxiety groups, which did not









face irrelevant peripheral cues, did not suffer any detrimental effects to the central driving

task.

Furthermore, in a previous study done using a similar dual task driving paradigm

(Murray & Janelle, 2000) to the one which will be used in this investigation, participants

performed under two experimental conditions: one in which they were instructed to

attend to the cues, and a second in which they were instructed to ignore the cues. The two

conditions were varied by task difficulty: low difficulty with reduced traffic (15 cars) and

computer driving aids; and high difficulty with increased traffic (30 cars) and removal of

computer driving aids. In the ignore-condition, participants produced significantly more

fixations for shorter duration than in the attend-condition, as well as an increase in

fixations to peripheral cues. Furthermore, slowest reaction times occurred during the

ignore condition on trials with invalid peripheral cues. Findings indicated that when

given instructions to ignore an attention-directing cue, the ability to ignore peripheral

cues is compromised if under relatively higher levels of mental load. The instruction to

ignore the cues in conjunction with increased stress by the increase in task difficulty

apparently led to the detriments in reaction time and an increase in saccadic eye

movement.

In summary, the available evidence strongly suggests that attention allocation

precedes saccadic shifts of attention, and that attention and saccadic movements are

coupled. Second, stress and anxiety can compete for sparse processing space, which

leads to detriments in previously learned or automatic processing. Third, changes in

attention allocation can lead to changes in fixation duration, and search rate. Thus,

diminishing processing space and the automatic process of attention allocation can









influence the processing efficiency of the individual. Furthermore, as task demands

increase, they consume more of the working memory capacity. In turn, resources of the

central executive will be pre-empted, making performance more vulnerable to anxiety

and interference from attentional demands placed on the performer.

Event Related Potentials (ERP) and Mental Effort

Eysenck and Calvo (1992) predicted that psychophysiological measures should be

able to detect changes in processing efficiency. There are, however, no known studies

which directly address the tenets of the processing efficiency theory through

psychophysiological indices. However, mental effort and attention have been addressed

through the use of electroencephalogram (EEG) (e.g., Fowler, 1994; Sommer, Matt, &

Leuthold, 1990; Stelmack, Houlihan, & McGarry-Robberts, 1993; Strayer & Kramer,

1990). The use of EEG can reflect changes in cognitive effort irrespective of changes in

performance. In addition, event-related brain potentials (ERPs), which are derived from

EEG, are often used to examine cognitive changes as they relate to specific events.

Before discussing ERPs and their relationship to mental load, a discussion of EEG and

ERP derivation is warranted. The section is organized in the following way: (1)

properties of EEG; (2) derivation and generation of ERPs; (3) sport related EEG research;

(4) P300 and mental load; and (5) summary.

Properties of EEG

The EEG reflects voltage variation from the summation of neuronal activity

originating within the brain recorded across time (Coles & Rugg, 1995). The EEG and

ERPs are recorded from electrodes placed on the scalp. The typical approach is to

arrange the electrodes according to the traditional 10-20 system (Jasper, 1958). This

system divides the scalp by the electrode's proximity to particular regions of the brain









(frontal, central, temporal, parietal, occipital) and by the lateral location (odd numbers to

the left, z denotes the midline, and even numbers are the right). Each lobe has been

linked to a specific process. High order functions such as emotion, language and motor

planning are represented in the frontal lobe. The central lobe is associated with motor

execution while the parietal lobes represent cognition and perception as well as sensory

motor function. Visual processing is dominant in the occipital lobe. Despite these

generalizations, it should be emphasized that the brain exhibits a high degree of

"plasticity", meaning that various cortical locations can be involved in a number of

different functions.

The voltage variation that makes up the EEG trace represents the difference in

voltage between two electrode sites. The most common practice is to employ a

'monopolar' common reference recording procedure (Coles & Rugg, 1995). The

common reference recording procedure involves connecting a single reference electrode

or a pair of reference electrodes (typically linked mastoids or linked ears) to each member

of an array of scalp electrodes. The scalp electrodes are active sites, whereas the

reference electrodes are inactive and relatively uninfluenced by electrical activity.

During the recording and amplification process, the common mode rejection (CMR) is

employed in which any common signal to the two recording sites (active and reference) is

considered non-cortical in origin. Using a differential amplifier, the non-cortical voltage

is automatically subtracted leaving only electrical activity emanating from the cortex.

EEG is measured in terms of amplitude and frequency, and is measured in

microvolts. Fluctuations occur with peak-to-peak amplitude smaller than 100 mV. Due

to the small amplitude, EEG is typically amplified with a gain factor from 20,000 to









50,000. Most EEG amplifiers have a frequency response from 0 Hz to 100 Hz or more.

However, the typical EEG recording involves frequencies between 0.5 and 50 Hz.

Frequencies above 50 Hz and below 0.5 Hz tend to be attenuated through filtering.

The EEG signal is converted to a digital signal using an analog-to-digital

converter, and inputted into a digital computer. The rate at which the single is sampled

be must determined to ensure the computer obtains an accurate record. Using the

Nyquist principle (Ray, 1990), an accurate record can be achieved. The Nyquist principle

states that a sampling rate must be at least twice the highest frequency in a given sample

to contain all the basic information in that signal. In practice, most EEG researchers take

a more conservative approach by using a sampling rate of at least three to five times the

highest frequency (Ray, 1990).

Once sampled, the EEG trace is then filtered from noise created by artifacts.

Artifacts are created from non-cortical activity such as the beating of the heart, movement

of the eyes or eyelids, muscular activity from the face or neck, as well as external

artifacts such as electric lights. The external factors are removed by using a band pass

filter that is set to remove all 60 Hz activity.

The major sources of endogenous artifacts are the movements of the eyes and the

eyelid activity. Filtering of eye and eyelid movement is not possible because these

movements occur at a similar frequency of the EEG waveform. Thus, electrooculogram

(EOG) activity is recorded along with the EEG. In investigation of only EEG

waveforms, EOG activity is minimized through instruction and through methodology.

However, in investigations of ERP waveforms, the researcher can identify ocular artifact,

and deal with these artifacts in one of two ways. The first method is to eliminate any









ERP in which eye blinks or eye movements are found. The problem with this method is

that data are lost, and if there is a significant amount of eye blinks or eye movements

(which will occur in this study), then there may be an insufficient number of artifact free

trials in which to draw inferences. The second method is to mathematically remove the

contribution of eye blinks and eye movements. In this second method, the ERP is kept in

tact and there is very little loss of needed data. Gratton, Coles, and Donchin (1983)

offered an unique approach to removing significant ocular artifacts in which a correction

factor is created from the EOG signal and is then applied to the raw ERP to create a

corrected averaged waveform. This approach is commonly used in ERP research today

and has been shown to be both valid and reliable (Brunia et al., 1989).

Once the filtering and removal of ocular activity is completed, the resultant EEG

is then grouped into various types of periodic activity by a process called spectral

analysis. The resultant frequencies are grouped and placed into traditional frequencies

bands: delta (1-4 Hz), theta (4-7 Hz), alpha (8-12 Hz), beta (13-35 Hz), and gamma (36-

44 Hz). Theta and delta bands are most commonly associated with sleep and represent

low levels of arousal. Alpha, beta, and gamma bands are representative of conscious

states. Alpha is most prominent during relaxation whereas beta and gamma bands are

most prominent during activation (Ray & Cole, 1985). However, Nunez (1995) offers a

more contemporary viewpoint on the relationship of alpha and beta waves in that these

waves provide an indication of communication within the cerebral cortex.

In sport, the use of EEG is becoming more common and is a viable instrument to

answer many important questions about the underlying mechanisms which effect

behavior. Hatfield, Landers, and Ray (1984) published one of the first studies examining









world-class competitive marksman during aiming, shot preparation, and shot execution.

Hatfield et al. (1984) found that left hemisphere (verbal/symbolic processing) activation

decreased during the sighting phase of the shot preparation, whereas right hemisphere

(visual/spatial processing) activation remained relatively high. They concluded that

highly skilled athletes decreased cortical interference by increasing the spatial

programming of the right hemisphere with a corresponding reduction of activity in the

left hemisphere. Since Hatfield et al.'s (1984) publication, novice and elite target

shooters have been a fruitful source to study mental states associated with motor

performance (e.g., Janelle et al., 2000; Konttinen & Lyytinen, 1992; Konttinen &

Lyytinen, 1993; Konttinen, Lyytinen, & Konttinen; 1995; Salazar et al., 1990). Other

researchers have used EEG to study motor performances that are similar to shooting, and

have generally found a similar reduction in left-hemispheric activation immediately prior

to and during performance. These include golf putting (Crews & Landers, 1993), archery

(Lander et al., 1994, Salazar et al., 1990), weight lifting (Gannon et al., 1992), and karate

(Collins, Powell, & Davies, 1990).

Another flourishing area of EEG research is its use in exercise. The primary goal

of exercise psychophysiological research is to understand the cortical mechanisms that

influence exercise or change as a result of exercise. Furthermore, exercise

psychophysiologists try to determine whether or not exercise causes changes in anxiety,

stress, and emotion as indicated by the brain. The typical approach has been to measure

EEG (primarily alpha and beta bands) immediately preceding and following a single bout

of exercise (e.g., Kubitz & Pothakos, 1997). Changes are seen through brain activation,

which is defined as a decrease in alpha activity with a corresponding increase in beta









activity across both hemispheres (e.g. Kubitz & Pothakos, 1997; Kubitz & Mott, 1996;

Petruzzello & Landers, 1994).

The opponent-process theory (Solomon, 1980), which proposes that the body's

goal has been to maintain homeostasis, is used to explain these changes. The opponent

process theory states that the body, when faced with an aversive stimulus (exercise), will

react with an opposite and corrective response. The brain is activated during exercise and

immediately following there is a decrease (below baseline) to initiate a return to

homeostasis, and a subsequent increase in alpha activity. Although a growing body of

literature on EEG and exercise exists, the bulk of this research in this area is replete with

methodological problems, lack of comparable results, and unsubstantiated conclusions.

Derivation and Generation of ERPs

When the voltage variation from the EEG is time-locked to a known stimulus and

averaged, it is possible to precisely examine the brain's response to the stimulus. The

averaged time-locked voltage variation is known as an event-related potential (ERP).

Though the voltage variation is time-locked to a known stimulus, the relationship

between the underlying process in the brain and what is observed at the scalp is not

completely understood. Considerations of the neural processes that are most likely

displayed and detected in the ERP have important consequences for their interpretation.

ERPs can be recorded using the same amplifiers and filters as used in EEG

recording (although, some systems are now being designed specially for ERP

recordings). The ERP is small (only a few microvolts) where as the EEG signal is large

(50 mV). So to extract the ERP waveform from the larger EEG, a process of averaging

must be employed. Averaging is a signal processing technique used to decrease the

signal (time-locked ERP) to noise (background EEG) ratio. To accomplish the averaging









process, a number of EEG epochs are recorded, which are time locked to the same event.

The time-locked EEG epochs are then averaged to yield a single waveform referred to as

the event-related potential. The basic assumption is that all of the EEG data that is not

time-locked to an event will randomly vary, and their average will tend to be zero leaving

only the ERP, or time-locked data left. The size of the signal to noise ratio will increase

by the square root of the number of trials included in the average (Ray & Coles, 1985). A

potential problem with averaging is the variation and variability in the ERP from trial to

trial and from individual to individual. However, with employment of a within subjects

design, this variation can be minimized.

The ERP is usually divided into components, which are the observable positive

and negative peaks that occur at some time just prior to or some latency following the

eliciting stimulus. The ERP is divided into exogenous and endogenous components.

Components that are elicited by the physical presence of a stimulus, and are considered

obligatory, are called sensory or exogenous. The exogenous components are elicited in

every trial and with every individual, unless the sensory system is compromised in some

fashion. Conversely, endogenous components are elicited by the interaction between the

stimulus and the individual. The endogenous components can be affected for example by

the degree of selective attention needed, task relevance, and by expected or unexpected

changes in the stimulus (Ray, 1990). Endogenous components are of most interest to

psychophysiologists due to the capability to manipulate their characteristics in

experimental settings. Though exogenous components can be affected by experiment

variation (e.g. attention), and endogenous components can be affected by sensory inputs

(e.g. modality of the stimulus), the exogenous-endogenous dimension can be thought of









as relative to time. Typically, exogenous components occur in the first 100 ms and

endogenous components occur following the first 100 ms (Coles & Rugg, 1995).

Components are typically defined as to when their latency occurs. For example, the

P300, which will be of interest in this study, is a positive wave, which typically occurs

approximately 300 ms after the eliciting stimulus. It, however, can occur anywhere from

250 ms to 800 ms, so in more recent years it has typically been referred to as P3.

The ERP provides a time-locked electrophysiological window of the sensory,

motor, and cognitive process of the brain, and even with the problems mentioned above,

ERPs can be used to understand cognitive processes and attentional mechanisms. The

typical approach to studying ERPs involves examines changes from when a certain

stimulus is present and when it is not. Thus, justification for the use of the ERP in

research on attention is three fold. First, ERPs provide a more complete picture of

cognitive processing than can be gathered from behavioral methods alone. Second, the

ERP provides a measure of attention irrespective of any requirement that the subject

attends to or responds to a stimulus (Mangun & Hillyard, 1995). Finally, due to the

temporal resolution of the ERP, important information can gained about the timing of

cognitive events that is difficult to infer from behavior (Mangun & Hillyard, 1995). A

brief review of some of the more commonly recorded event-following components

follows.

As stated previously, event following components can be divided into exogenous

and endogenous components. Exogenous or sensory components are a series of

deflections in the ERP that are associated with all modalities and are considered

obligatory. Deflections are caused by the transmission of sensory information, and occur









up to 100 ms following the eliciting stimulus. Following the sensory components, the N2

(also referred to as mismatch negativity) occurs at about 200 ms following the

presentation of either a visual or an auditory event. A large N2 is elicited for rare stimuli

regardless of task relevance. To detect the N2, it is standard to subtract the ERP for a

frequently occurring stimulus from the ERP of a rare stimulus (Mangun & Hillyard,

1995). The result is the proportional difference between rare and frequent stimuli (i.e. the

mismatch negativity). N2 represents an orienting reflex to events that deviate from the

prevailing context. In other words, the N2 amplitude is related to the probability that an

event will occur (i.e., the less likely the event, the larger the N2).

The P3 or P300, which has received more attention than any other component,

follows the N2. The probable explanation for the prevalence of research employing the

P3 is the relative ease with which it can be elicited as well as its size (5-20 mV) (Coles &

Rugg, 1995). The P3 was originally elicited in an 'oddball paradigm' in which two

classes of stimulus are presented. The probability of one stimulus occurring is

considerably greater than the other. During the presentation of the rare stimulus, a larger

P3 is elicited. The latency variability of the P3 in the oddball paradigm is controlled by

the ease at which the rarer stimulus can be identified. The P3 wave is also elicited during

the occurrence of a task-related event (Coles, Gratton, & Fabiani, 1990), and these task-

related events can either effect the latency, amplitude, or both of the P3 waveform.

In studies of selective attention, the general approach to elicit the P3 is through

the dual task paradigm. The P3 amplitude can be affected by the amount of attention that

is applied to a certain stimuli, and is viewed as a measure of processing in a limited

capacity system (Donchin & Coles, 1988). When two strings of stimuli are presented, the









P3 amplitude will be larger for the attended stimulus (Picton, 1992). Furthermore, the

amplitude of the P3 is decreased by an increase in the perceptual demands of the

secondary task (Nash & Fernandez, 1996). The P3 is not affected by increased motor

difficulty of the secondary task because the perceptual and motor resources are separate,

and the P3 is primarily related to perceptual resources (Picton, 1992).

The latency of the P3 is considered to provide a measure of the timing of the

evaluation process (Coles et al., 1995). P3 latency increases as discrimination or

categorization of stimuli becomes more difficult (Kutas, McCarthy, & Donchin, 1977).

Accuracy, which is dependent on discrimination of the stimulus, also effects latency.

More difficultly associated with the discrimination of a stimulus decreases the odds that

the stimulus will be identified correctly, thus increasing the latency of the P3. For

example, using a choice reaction time task, Coles, Gratton, Bashore, Eriksen, & Donchin

(1985) demonstrated that shorter P3 latencies were produced when accuracy was higher.

Coles et al. (1985) proposed that the short P3 latency indicate that the evaluation process

has proceeded quickly. However, in situations of longer P3 latency, more information or

evaluation is required before proceeding with the motor response. Finally, P3 latency is

not affected by stimulus-response incompatibility, which typically has a large influence

on reaction time (Coles et al., 1995).

Finally, N400 is primarily related to meaning and emotionality of words. Kutas

and Hillyard (1980) first observed the N400, which is a negative deflection with a latency

of about 400 ms, in studies of sentence structure and processing. In their study, seven

word sentences were presented with each word being presented individually followed by

a 1 second interval. The final word in each sentence presented was either semantically









correct or semantically incorrect. The semantically incorrect words elicited the N400.

Thus, the N400 appears to be related to the semantic relatedness of words and the

sentence context, and is largest when words meaning is not predicted by or unrelated to

the prior context or the words.

In summary, this section represented an overview of the more common ERP

components in psychophysiological research, and a more in depth discussion is not

warranted in this paper. Generally, it can concluded that N2 indexes rare stimuli

regardless of task relevance; the P3 represents the amount of attention allocated to

stimulus discrimination, and the N4 represents the meaning and emotionality of words.

For a more complete review, see Rugg and Coles (1995).

P3 Amplitude and Mental Effort

The proceeding section offered an overview of some of the issues addressed in

ERP research. This section highlights the use of the P3 in dual task situations and

presents empirical evidence to support P3 appropriateness for use in testing the

predictions of the PET. Evidence suggests that P3 is an acceptable measure of the

processing efficiency as it can be modulated by task difficulty and attentional allocation.

In this regard, it becomes an important measure for determining the cognitive processing

changes used in this investigation as they relate to the PET. In fact, one could contend

that investigating variability in the amplitude of the P3 is the most direct way to examine

the core notions of the PET.

A growing body of literature indicates that the P300 amplitude provides an index

of the allocation of cognitive processing resources in dual task paradigms (e.g., Isreal,

Chesney, Wickens, & Donchin, 1980; Sirevaag, Krames, Coles, & Donchin, 1989;

Wickens, Kramer, Vanessa, & Donchin, 1983). As stated previously, the PET proposes









that high anxious individuals apply more cognitive resources to the primary task in order

to maintain or improve performance. In essence, high anxious individuals will have less

processing space available for the secondary task, either through an increase in effort on

the primary task, or through a reduction in cognitive space brought on by anxiety. Thus,

it is expected that for high anxious individuals, the P3 amplitude will decrease for the

secondary task when compared to low anxious individuals.

Wickens, Kramer, Vanessa, and Donchin (1983) and Sirevaag et al. (1989)

examined P3 amplitude changes in both a pursuit-tracking task (the primary task) and to

the response of occasional auditory tones (the secondary task). Task difficulty was

manipulated by varying the control dynamics and the predictability of position changes of

the target in the primary task. As the difficulty of the tracking task increased, the P3

elicited by the primary task increased in amplitude. Conversely, the P3 amplitude for the

secondary task decreased during an increase in primary task difficulty. This P3

amplitude effect is known as P3 amplitude reciprocity. The P3 amplitude reciprocity is

interpreted as a trade-off in resource demands between the primary and secondary tasks.

Thus, the assumption is that an increase in difficulty in the primary task requires the

additional allocation of resources to the primary task thereby decreasing or depleting the

processing for the secondary task stimuli (Kramer, Trejo, & Humphrey, 1995).

More recent evidence was provided by Nash and Fernandez (1996) who

demonstrated and confirmed that the P3 can serve as an index of the distribution of

attentional resources. In addition, they also considered whether the timing of the

presentation of stimuli would affect the P3 amplitude. They presented an auditory

primary task and a visual stimuli secondary task which were separated by an interval of









400 ms. Participants were either placed in an attend condition where they were required

to count the presentation of deviant and standard tones, or they were placed in an

'unattend' condition where the auditory tones were to be ignored. Following the

presentation of the auditory tone (standard or deviant) and the delay interval, standard or

deviant LEDs were presented where participants in both conditions were to respond to

the deviant LEDs. The attend condition produced a larger P3 amplitude for the auditory

task, and a subsequent reduction in P3 amplitude for the secondary or visual stimuli task.

These results provided evidence that the P3 is an index of neural activity associated with

the reallocation of limited processing resources, and the processing of specific stimulus

information. The authors also indicated that the P3 enlargement for the primary task also

represented an inhibitory mechanism for response to the visual stimuli. In other words,

the increase in attentional allocation for the first stimulus inhibits the shift of attention

and elicitation of the P3 for the second stimulus.

In summary, the P3 amplitude reciprocity effect appears to be a well-documented

and well-supported phenomenon. However, there is still some debate about the specific

nature of the cognitive resources involved. The basis for this debate is related to

questions surrounding filter or resource models of attention. As stated earlier, there is a

lack of agreement among researchers (e.g., Donchin & Coles, 1988; Verleger, 1988)

regarding which attentional theory best represents cognitive processing and the processes

associated with the P3. Beyond the argument about the processes associated with the P3,

the evidence presented demonstrates that the P3 manifests the activity of a limited

capacity system, which remains constant even when presented with multiple tasks. In

reference to the PET, which also assumes a limited capacity system, the P3 acts as a









marker for the efficiency and effectiveness distinction. It is therefore expected that as an

individual applies more effort (or allocation of attention) to the primary task, the P3

amplitude for the secondary task will decrease.

Summary

In this chapter, literature describing current theoretical models that explain the

anxiety-performance relationship was reviewed, and support was provided for the use of

the processing efficiency model in sport-related research. While classic anxiety-

performance models, such as the inverted-U hypothesis and the drive theory, represented

guiding first steps in understanding the anxiety-performance relationship, their

prevalence in recent research has greatly diminished due to the lack of a

multidimensional representation of anxiety. The more recent multidimensional anxiety

theory represents anxiety as a multidimensional property, but support for it has been

equivocal due to evidence that demonstrates anxiety as having both facilitative and

debilitative properties. The cusp catastrophe model and the processing efficiency theory

represent two attempts to explain the facilitative and debilitative effects of anxiety.

However, because the cusp catastrophe model is not a theory, it does not offer any

explanation of the underlying mechanisms for these differential effects, and is based on

global performance measures that might not be sensitive to cognitive changes caused by

anxiety. Thus, the processing efficiency theory appears to be most appealing in that it

represents an explanation of the underlying processes, which influence anxiety.

However, since the cusp catastrophe model and the PET represent complimentary

explanations rather than competing ones, there is a potential that both explanations could

be jointly developed further. The processing efficiency theory proposes that anxiety

affects processing efficiency to a greater extent than performance effectiveness and that









impaired processing efficiency can be improved through an increase in additional

processing resources for the task leading to increased or maintained performance.

Current research on the processing efficiency theory provides only elementary

support for the conclusion that increased resource allocation to a specific task will occur

in order to maintain performance of that performance of that task. There is a need for a

more complex and direct account of the predictions of this theory. In addition, the

majority of research completed to date has employed tasks with a substantial verbal

component. Eysenck (1996) suggested a need to test the theory on motor tasks. With this

mission as the underlying motivation for the current investigation, the central tenets of

the processing efficiency theory in a dynamic situation will be evaluated using not only

response time, but also ERPs and visual search as measures of the processing efficiency

and performance effectiveness distinction. The following chapter describes the

methodology used to evaluate the PET in a dynamic situation.














CHAPTER 3
METHODS

This chapter describes the methodology that was used to empirically test the

processing efficiency theory in a dynamic simulated sport situation. The empirical test

and subsequent analysis relied on data collected during a series of tasks performed on a

driving simulator. Variables of interest and importance included those designed to

measure the degree to which participants experience anxiety as well as measures of

performance effectiveness and processing efficiency.

Participants

In order to insure an adequate sample, a power analysis was conducted using an

effect size of 0.65 (based on response time pilot data) with power set at 0.8 and alpha set

at .05. Thus, 28 male volunteers were randomly recruited from various elective

university sport and fitness classes, and received extra course credit for participation.

These students were assigned to one of two condition groups identified as the: 1) lower

trait anxiety group, or 2) higher trait anxiety group. Trait anxiety was determined with

scores from the STAI (Spielberger, 1983), and groups were formed based on the

normative data, in which college males averaged 38.30 (SD = 9.18). Thus, the higher

trait anxious group scored one-half a standard deviation above the mean or higher, and

the lower trait anxious group scored one-half a standard deviation below the mean or

lower. Overall, forty-six males completed the STAI, and eighteen were eliminated

because they did not match the preset criterion for inclusion in the study.









Tasks, Instruments, and Objectives

A dual-task paradigm (i.e., a driving simulator and a response time task) was used

to evaluate the processing efficiency theory. Participants were told that performance on

both tasks was equally important, and that their goal was to try to avoid letting the one

task effect performance on the other. In addition, they were informed that their outcome

scores would be based on total performance on both tasks. That is, their performance on

both the driving and response time tasks were combined for use in the competition

session (Janelle et al., 1999). All testing was conducted in a dimly lighted room to

reduce distraction and to allow for proper testing. Each participant was seated 2.0 m

from a 1.72 m x 1.57 m screen.

Primary task. The primary task was driving an IndyCar racing simulator that

required the participants to navigate the Michigan racecourse. The simulator consisted of

racing software, an analog steering wheel, brake and accelerator foot pedals, and a

projection unit. The computer software used with the simulator is the Papyrus Design

Group IndyCar Racing II CD (Belleview, WA). The Michigan racecourse was projected

onto the screen using a Sharp Notevision Projector (Model # XG-NV2U, Camas, WA)

from a Gateway 2000 Solo 2 233 MHz Pentium computer (Sioux City, SD). The driving

task difficulty was deliberately minimized through a reduction of traffic on the road and

through computer steering controls. This afforded participants the opportunity for

improvement, and diminished the potential of confounding variables such as differences

in road conditions for each session, which might affect performance.

Secondary Task. The secondary task was a response time (RT) task related to the

appearance of one of four peripherally located light emitting diodes (LED) on the









projection screen. The secondary task was based on the relationship between the cost of

making a mistake versus the benefit derived from performing the task correctly. In this

task, central and peripheral arrows (i.e., cues) either correctly or incorrectly directed the

participant to the appropriate target.

A warning light, which was a centrally located LED in the participants' field of

vision, initiated the response time trial and was followed by one of three cues: a valid

cue, an invalid cue or a neutral cue. Valid and invalid cues were one of four arrows

pointing either left or right. These cue arrows were illuminated in either central or

peripheral locations on the projection screen whereas a single LED (the warning LED),

located in the middle of the screen, served as the neutral or non-directional cue (see

Figure 3.1). A valid cue directed the participant in the correct direction to the target and

an invalid cue incorrectly directed the participant to the wrong target area. Seventy

percent of the trials were valid trials where a cue arrow correctly indicated the position of

the target (Jonides, 1981). Twenty percent of the trials were invalid trials where the cue

arrow incorrectly indicated the target location. The remaining 10% of the trials were

neutral trials where the warning light was repeated and no direction was given to indicate

the location of a target (Jonides, 1981).

While attending to the driving task, participants were also required to press one of

two buttons dependent on which cue was illuminated. Participants were instructed to

press the right button if the top light was illuminated on either side of the screen, and to

press the left button if the bottom light was illuminated on either side. The buttons were

mounted on the steering wheel in convenient positions. Specifically, the left button was

located at 300 degrees and the right button at 60 degrees relative to the top of the steering









wheel. The onset of the warning light, as well as precues and cues, was controlled by a

computer program written on a 486 IBM PS/Value Point PC created specifically for this

task. The computer program allowed for the control of duration of illumination of all

lights, duration of delays between light presentation, and the percentage of valid, invalid

and neutral trials. The warning light and the precue arrows contained 5 mm in diameter

LEDs, while the peripheral cues contained 4-mm diameter LEDs. Thus, the small size of

the peripheral cues required a goal directed saccade in order to discriminate whether the

top or the bottom LED was illuminated.







Directional Directional Directional Directional
Cue Cue Cue Cue



Target Warning Target
&
Neutral Cue



-4 Simulation -
Viewing area


Figure 3.1. Experimental setup.

Measurement Devices and Performance Measures

Several measures were employed to examine processing efficiency and control of

attention. Response time, visual search, and an electroencephalogram (EEG) were each

used to measure processing efficiency, while driving performance in the form of lap









speed and number of laps completed was the indicator of performance effectiveness.

State anxiety was measured by the Competitive State Anxiety Inventory-2 (CSAI-2:

Martens et al., 1990) whereas the State-trait Anxiety Inventory (Spielberger et al., 1970)

was used to measure trait anxiety. Through the use of the Biopac MP100 system (Biopac

Systems Inc., Santa Barbara, CA), it was possible to record the presentation of lights, the

participants' reaction, and EEG concurrently. Response time data and saccadic

movement data were also integrated through the Biopac MP100 system. This integration

permitted time-locked evaluation of the data. Dependent measures included:

1) Driving task performance. Driving task performance was measured by the

number of laps completed in two 5-minute driving tests per session and reported as the

average speed per lap.

2) Response time task performance. Response time was measured using

Acqknowledge Software (Biopac Systems Inc., Santa Barbara, CA) and consisted of the

duration from the onset of the cue LED and the depression of the steering mounted

response button. Both correct and incorrect responses were recorded, but only correct

responses were analyzed.

3) Visual search. Eye movements were recorded with an Applied Science

Laboratories (ASL, Waltham, MA) 5000 SU eye movement system with Eyehead

Integration software. The 5000 SU eye movement system is a video based monocular

corneal reflection system that accurately measures line of gaze with respect to the

orientation of the head. Eyehead integration allowed for the digital recording of eye

movements by the integration of eye and head position with respect to a fixed scene

space. The scene space was defined as a "set of planes, consisting of one calibration









plane and a variable number of additional, bounded planes" (ASL, 1997).

The system has an accuracy of 10 visual angle with precision of 0.50. The eye

movement system consisted of a visor, an eye camera, a scene camera, an eye tracker

control unit, a head tracker electronics unit, a head mounted magnetic sensor, a magnetic

transmitter, an eye monitor, a scene monitor, and a 133 MHz Pentium PC. The eye

camera recorded the cornea reflex and pupil size of the right eye, which was then used to

compute both pupil diameter and line of gaze. All eye movement data were recorded on

the 133-MHz Pentium PC and analyzed through the Eyenal program (ASL, Waltham,

MA). The Eyenal software allowed for calculation of eye movements, fixation location,

fixation duration, and scan patterns using Area of Interest (AOI) files. AOI files define

scene areas with meaningful coordinates. There were seven AOI files, which matched

the location of the warning light, precue arrows, and cue lights. The seven AOI files

were used for an analysis of fixation duration and number of fixations at each point of

interest. Furthermore, the eye tracking system was also utilized to establish search rate,

which refers to the number of fixations, and the duration of each fixation during task

performance (Williams et al., 1994). Saccade latency was established from the electro-

oculographic activity as recorded with the Biopac system. Saccade latency was defined

as the time from the moment of target presentation to the initiation of the saccadic

movement.

4) EEG Recording. The electroencephalogram (EEG) was recorded from

nonpolarizable silver-silver chloride (Ag/AgC1) electrodes mounted in an ElectroCap

International elastic cap. Three electrode sites (Fz, Cz, and Pz) based on the International

10-20 system (Jasper, 1958) were prepared using Omni-prep and ECI electrode gel. All









sites were referenced to linked mastoids. The bipolar electro-oculographic activity

(EOG) was recorded to monitor vertical and horizontal eye movement artifacts. All

ocular artifacts were corrected using the procedure described by Gratton, Coles, and

Donchin (1983). The electrode impedance for EOG and EEG electrodes was less than

five kohms. The signals were amplified 50,000 times by a Biopac Systems amplifier

with a .01-100 Hz bandpass filter and digitized at 256 Hz using a Biopac Analog/Digital

converter, and recorded on-line using Acknowledge 3.2.6 software installed on a 233

MHz Gateway 2000 Pentium computer.

5) Cognitive anxiety. The CSAI-2 (see Appendix A) provided data to assess the

situational level of cognitive anxiety. The CSAI-2 is a valid and reliable instrument that

measures cognitive anxiety, somatic anxiety, and self-confidence. The scale contains 27

items with 9 items in each of the three subscales. An individual's intensity of their

anxiety symptoms was evaluated by a 4-point Likert scale (1 = not at all to 4 = very much

so). The CSAI-2 was chosen because it is a sport specific measure of competitive state

anxiety.

6) Trait Anxiety. Trait anxiety was determined by the STAI. The STAI is a 20-

item questionnaire that has both a trait and a state form. A 4-point Likert scale (1 = not at

all to 4 = very much so) indicated an individual's trait level of anxiety on each item.

7) Heart Rate. Heart rate, which is an indicator of arousal, was measured using a

Biopac electrocardiogram amplifier. Nonpolarizable silver-silver chloride (Ag/AgC1)

electrodes were placed on the interior of the left and right wrist, and the ground electrode

was placed on the bony portion of the ankle. The three electrodes were prepared using

Omni-prep and ECI electrode gel. The electrode impedance for ECG electrodes was less









than five kohms. The signals were amplified 500 times by a Biopac Systems amplifier

with a 1.0 -100.0 Hz bandpass filter and digitized at 256 Hz using a Biopac

Analog/Digital converter, and recorded on-line using Acqknowledge 3.2.6 software

Procedure

Participants were pre-screened with the STAI, and the video game questionnaire

(see Appendix B). Only participants who scored above 43 and below 33 on the STAI and

score 9 or lower on the video game questionnaire were included in the study. The

participants were seated comfortably in front of the viewing screen in a dimly lit room

with the steering wheel and the foot pedals comfortably positioned. After completing the

informed consent form (see Appendix C), the participants were given general instruction

on the purpose of the study, and were shown the data collection apparatus. Before the

application of testing equipment, participants were given the opportunity to gain

familiarity with the driving task by practicing the tasks until they attained an average lap

speed of 200 mph for at least 3 consecutive laps performance level. For the purpose of

this investigation, performance was of interest rather than learning. Therefore, attaining

this performance level helped to reduce learning effects that might have otherwise

confounded the data.

Following the Society for Psychophysiological Research guidelines (Putnam,

Johnson, & Roth, 1992), the participants were prepared for electrocortical measurement,

and were fitted with the appropriate Lycra electrode cap. Following impedance testing

and EEG calibration, the eye-tracking headband was placed on the participants' head, and

then calibrated. The calibration involved using a nine-point reference grid that

represented known positions. A reference grid was generated by Microsoft PowerPoint









97 and projected onto the screen. The participant fixated on each point in the nine-point

grid and each fixation was entered into the computer memory, thus creating a match

between the location of the individual's eye and the known location of the nine points on

the reference grid. After completion of the eye calibration, electrodes to measure heart

rate were placed on the participants' wrists and a bony portion of their right ankle.

Two test sessions were completed on separate days. The initial session included

instruction on the function of the driving task and the response time task. Immediately

before the initiation of testing, all participants were asked to complete the CSAI-2 in both

sessions. Following the completion of Session 1, participants were shown their average

lap speed during the first testing session and instructed that they were placed in the

competitive group according to the statements in Appendix D. Preceding Session 2, the

participants received reminders about how to perform the task and the experimental

manipulation (see Appendix D). First, the participants were given the ego-threatening

instructional set followed by the motivational instructional set (see Appendix D). Prior to

performance, the participant completed the CSAI-2. The driving task was a 10-min

timed driving task divided into two 5-min blocks where the goal was to complete as

many laps as fast as possible. The participants were told that their driving performance

was based on the number of laps they completed and the average lap speed.

Concurrently, participants were required to attend to the response time task. A

response time trial began with the onset of the warning light for 250 ms followed by a

delay that randomly varied between 150 to 750 ms. The onset of one of the four arrows

(i.e., valid or invalid precues) or the warning light (i.e. the neutral precue) for 250 ms

followed this delay. A delay between 90 and 150 ms preceded the onset of the target









light, which was illuminated until either the participant depressed the response time

button or 3 s had elapsed. Participants were instructed to react as quickly and accurately

as possible. There was a 3 s delay between response time trials. The total duration for a

response time trial including the inter-trial delay was between 2.75 s and 3.40 s for a total

of 88 to 110 valid, invalid, and neutral response time trials for each 5-min session. The

minimum number of trials per cue type for averaging was 30 (Picton, 1992).

ERP, Heart Rate, and Eye Movement Data Reduction

EEG data reduction occurred offline using Acqknowledge 3.2.6 software. Data

were filtered with a bandpass of 0.01-30 Hz, and the ERP and EOG was sampled from

100 ms prior to cue onset to 900 ms post-cue onset. The amplitude of the P3 component

was defined as the highest peak in the interval, 250-750 ms following the onset of the

directional cue, and was measured with respect to the mean of the 100 ms pre-cue

baseline.

Since the methodology requires significant yet predictable eye movement, ocular

artifacts were removed using an eye movement correction procedure by Gratton, Coles,

and Donchin (1983). The procedure was as follows: (1) A raw average was created for

each condition from all trials and from each electrode. (2) The raw averages were then

subtracted from the single trial record and were considered an estimate of activity at a

given electrode site that was not event-related. (3) The EEG data from step 2 was no

longer time locked and was considered a dependent variable with the EOG data as the

independent variable in a regression equation. Through the least-squares technique, it

was possible to estimate the correction factor (K), which was calculated separately for

blinks and eye movements. (4) The raw EEG data were then subtracted from the EOG









value scaled by the correction factor (K). (5) Following subtraction of the scaled EOG,

the data was averaged to yield a corrected time locked ERP.

Heart rate was established using the Acqknowledge 3.2.6 software, which

includes an R-wave function. The R-wave function reduces any components in the

waveform that might be mistaken for peaks. The R-wave function reduces the waveform

through 17 Hz bandpass filter followed by a full wave rectifier, and a 10 Hz low pass

filter. The resultant waveforms supplied information to calculate beats per minute across

each session.

Visual search data were reduced using Eyenal software. The Eyenal software

reduces raw eye position data to a series of fixation points, and allows for the calculation

of mean fixation duration, number of fixations, fixation frequency, and saccade length.

Furthermore, each of these measures was referenced to the established AOI files.

Saccade latency was calculated with information from the Biopac MP100 system, and

reduced using Microsoft Excel (Microsoft Corporation, Roselle, IL).

Design and Analysis

To test the hypotheses of interest, the following statistical analyses were applied.

Alpha was set at p = .05 for all statistical tests. For comparisons that required analyses of

variance (ANOVA), Tukey's HSD post-hoc analysis was used when necessary to

evaluate significant main effects, whereas simple effects tests were conducted as follow-

up analyses for significant interactions. Also, when necessary, violations of the

sphericity assumption were corrected using Greenhouse-Geisser adjustments of the

degrees of freedom. For multivariate analyses of variance (MANOVA), significant main






75


effects and interactions were evaluated through follow-up univariate analysis of variance

(ANOVA) tests.














CHAPTER 4
RESULTS

Results are presented in the following order (1) cognitive anxiety and heart rate,

(2) dual task performance, (3) visual search measure, (4) ERP measure, and (5) the post-

experiment questionnaire data. The symbols for the following tables, charts, and graphs

are as follows: LA lower trait anxiety, HA higher trait anxiety, IP invalid peripheral

cue, VP valid peripheral cue, IC invalid central cue, VC valid central cue, and N -

neutral cue.

Pretest Performance Criterion

To ensure equal groups, the length of time to achieve the criterion performance

level was analyzed using a t-test (t (26) = 1.08), which produced non-significant results.

Overall, the participants required 18 to 35 min to reach the criteria average lap speed

level. The mean duration for the HA group was 27.78 (SD = 3.42) min, whereas the mean

duration for the LA group was 26.07 min (SD = 4.84).

Trait Anxiety

Participants were assigned to one of two groups identified as the: 1) lower trait

anxiety group, or 2) higher trait anxiety group. Trait anxiety was determined with scores

from the STAI, and groups were formed based on normative data (Spielberger, 1983) in

which college males averaged 38.30 (SD = 9.18). The mean for the higher trait anxious

group was 44.14 (SD = 1.35), while the mean for lower trait anxious group was 27.85

(SD = 2.93).









Cognitive Anxiety and Physiological Arousal

To determine the effect of changes in cognitive anxiety as measured with the

CSAI-2, and physiological arousal as determined by heart rate, a 2 (Group) x 2 (Session)

multivariate analysis of variance (MANOVA) with repeated measures on the last factor

was employed. This test revealed significant main effects for Group (Wilks' Lambda =

.677, F(2, 25) = 5.96, p < .001) and for Session (Wilks' Lambda = .095, F(2, 25) =

119.46, p < .01). In addition, and more importantly, the analyses indicated a significant

Group x Session interaction (Wilks' Lambda = .667, F(2, 25) = 6.24, p < .01). Follow-up

2 (Group) x 2 (Session) univariate analyses of variance with repeated measures on the

last factor were conducted on heart rate and on cognitive anxiety. Analysis of heart rate

data indicated a significant Session main effect, F(1, 26) = 17.02, p < .01. Both groups

demonstrated an increase in heart rate from Session 1 M = 84.17, SD = 1.91) to Session

2 (M = 96.01, SD = 3.99).

The univariate ANOVA for cognitive anxiety indicated a significant main effect

for Session, F(1, 26) = 184.83, p < .001, and for Group, F(1, 26) = 10.81, p < .01. A

significant Group x Session interaction, F(1, 26) = 10.36, p < .01 was also found. Simple

effects analysis revealed that both groups increased significantly from Session 1 (HA: M

= 9.57, SD = 0.51; LA: M = 9.42, SD = 0.52) to Session 2 (HA: M = 22.07, SD = 4.61;

LA: M = 17.30, SD = 3.21). However, the higher anxiety group demonstrated a

significantly larger increase than did the lower anxiety group (See Figure 4.1).






78



25

20

S15
--- HA
.> -- LA
t 10

5

0
1 2
Session


Figure 4.1. Changes in cognitive anxiety for each group across sessions.

Overall, the results from the CSAI-2 and HR revealed an increase in anxiety from

Session 1 to Session 2, and that the experimental manipulations produced a larger

increase in cognitive anxiety for higher trait anxious participants than for lower trait

anxious participants.

Dual-task Performance

Task performance was evaluated through performance on the driving simulator

and response time to the peripheral targets. To assess changes in driving performance,

average lap speed was calculated for the each of the sessions. Response time represented

the duration of time from target onset to button depression.

Driving performance. Driving performance was assessed using a 2 (Group) x 2

(Session) mixed model ANOVA with repeated measures on the last factor. The ANOVA

for lap speed revealed no significant main effects or interactions. The HA group's

average lap speed was 146.85 mph (236.28 km/hr) (SD = 11.42) in Session 1 and 155.80









mph (263.33 km/hr) (SD = 10.95) in Session 2 whereas the LA group's average lap speed

was 153.54 mph (259.51 km/hr) (SD = 11.98) in Session 1 and 156.58 mph (252 km/hr)

(SD = 15.22) in Session 2.

Response time. A baseline response time test was conducted where the

participants only performed the secondary task. A 2 (Group) x 5 (Validity) ANOVA was

conducted in order to test the effects of cue validity. The results revealed a significant

main effect for Validity, F(4, 104) = 20.76, p<.001. Tukey's post-hoc test exposed

significant differences between invalid locations and valid locations, as well as valid

locations and neutral locations (see Table 4.1). These results demonstrated the costs

associated with invalid cues, the benefits of valid cues, and the relatively moderate effect

of neutral cues. Curiously, the cost of invalid cues was not affected by location (central

or peripheral) in that there was very little difference between these cues. The same

results occurred for valid cues.

Table 4.1

Average Baseline RT


Validity M SD

IP .73 .02
VP .60 .01
IC .71 .02
VC .61 .01
N .68 .01


In the two dual task experimental sessions, a 2 (Group) x 5 (Validity) x 2

(Session) mixed model ANOVA was utilized to test the effects of the cost and benefit

paradigm on response time. This analysis indicated a main effect for Validity, F(4, 104)









= 64.71, p < .001 (see Table 4.2), and Group, F(1, 26) = 5.32, p < .05. Tukey's post-hoc

test for validity revealed that each of the cue locations was significantly different from

each other (see Table 4.2). The HA group displayed longer RT (M = .84, SD = .2) when

compared to the LA group (M = .77, SD = .2)

Table 4.2

Mean RT as a Function of Cue Validity


Validity M SD

IP .95 .02
VP .66 .02
IC .85 .01
VC .74 .02
N .80 .02


A more important finding was the Group x Session interaction. A simple effects

test revealed that groups were similar in Session 1, but significantly different in Session

2. Furthermore, the simple effects test demonstrated a significant reduction of RT from

Session 1 to Session 2 for the LA group, and conversely, a significant increase of RT

from Session 1 to Session 2 for the HA group (see Figure 4.2).

Overall, RT results across sessions indicated a decrease in secondary task

performance for the HA group with increase in performance for the LA group, although

participants in both groups were able to maintain performance on the primary task.

Visual Search Data

To describe alterations in visual search strategies, measures including fixation

location, fixation duration, frequency of saccades to central and peripheral locations,

search rate, and saccade latency were employed. After a careful examination of the raw









gaze location data from the ASL 5000 eye tracking system, and due to the large number

of zero values in many of the AOI files for each subject, the seven AOI files were

reduced to 2 locations: peripheral or central.


0.8 8
0.86
0 .8 4
0 .8 2
0.8 -- HA
0.78 LA
0.76 -
0 .74
0 .72 -
0.7
1 2
Session


Figure 4.2. Mean RT for each group across sessions.

The four peripheral locations (2 peripheral cues and 2 peripheral targets) were combined

to create one peripheral location, and the three central locations (neutral cue and 2 central

cues) were combined to create one central location. Search rate was calculated by

dividing the total number of fixations by fixation duration across all fixation locations.

For a saccade to be registered, a fixation needs to be maintained for a minimum period of

100 ms during which the eye does not move more than one degree. Saccade latency was

calculated as the duration of time from target onset to initial saccadic movement toward

the cue.

Fixation duration and location analysis. A 2 (Group) x 2 (Location) x 2 (Session)

MANOVA with repeated measures on the last factor was used to address fixation









duration and number of fixations at each fixation location. The analysis revealed a

significant main effect for Session (Wilks' Lambda = .68, F(2, 25) = 5.72, p < .01), and

for Location (Wilks' Lambda = .022, F(2, 25) = 599.61, p < .001). Follow-up separate 2

(Group) x 2 (Location) x 2 (Session) univariate analyses of variance with repeated

measures on the last factor were conducted on fixation location and fixation location

duration.

The univariate analysis of fixation duration revealed a significant Session main

effect, F(1, 26) = 8.86, p < .01, and a significant main effect for Location, F(1, 26) =

226.23, p < .001. Both groups demonstrated a decrease in the overall duration of

fixations in the specified areas from Session 1 (M = .33, SD = .007) to Session 2 (M =

.31, SD = 0.006). The significant main effect for Location revealed the expected result of

longer fixations in central locations (M = .41, SD = 0.006) as opposed to the peripheral

locations (M = .22, SD = .01). A most interesting finding is that participants decreased

their overall fixation duration in both areas of interest from Session 1 to Session 2.

The univariate ANOVA for fixation location indicated a significant main effect

for Location, F(1, 26) = 1016.04, p < .001 in that the vast number of fixations were

located in central area (M = 450.46, SD = 13.41) rather than the peripheral locations (M =

32.59, SD = 3.89). The results revealed that participants exhibited more fixations to the

central area as opposed to the peripheral, which was to be expected.

Search rate data analysis. A 2 (Group) x 2 (Session) ANOVA with repeated

measures on the last factor was used to evaluate changes in search rate. A Session main

effect (F(1, 26) = 4.73, p < .05) was revealed with an overall increase in search rate from

Session 1 (M = .39, SD = .015) to Session 2 (M = .33, SD = .020). Figure 4.3 provides










an illustration of the results. The results of these analyses provide evidence that during

Session 2 both groups had a greater number of fixations for shorter durations than in

Session 1.


0.41
0.4
0.39
0.38
f 0.37
SSession 1
o 0.36
U, Session 2
O 0.35
0.34
0.33 -
0.32
0.31
HA LA
Group


Figure 4.3. Mean search rate for each session by group.

Saccade latency. To determine the effects of saccade latency, a 2 (Group) x 5

(Validity) x 2 (Session) mixed model ANOVA with repeated measures on the last factor

was conducted. Significant main effects were found for Validity (F(4, 26) = 64.98, P <

.001), Group (F(4, 26) = 64.98, p < .001) and Session (F(1, 26) = 6.77, p < .05).

However, more informative were the two significant interactions: (1) a Group x Validity

interaction (F(4, 26) = 20.17, p < .001); and (2) a Validity x Session interaction (F(4, 26)

= 5.50, p < .01). Simple effects tests for the Group x Validity interaction (see Figure 4.4)

revealed significant differences between groups for IP cues, IC cues, and VC cues. For

the LA group (See Table 4.3), the saccade latency represents an expected pattern in

which valid cues resulted in an attentional enhancement to peripheral cues and invalid

cues represented an attentional cost. In other words, the saccade latency demonstrated a








cost and a benefit based on cue validity. Aside from the VP cue, the HA group had a

reduction in the influence of cue validity. Rather than saccade latency changing with

each cue validity, the HA group had an increase in saccade latency for IP, IC and VC

cues, which was not expected.

Table 4.3

Saccadic Latency Durations


Validity

IP
VP
IC
VC
N


0 .600

0 .500

0 .4 0 0

0 .300

0200

0 .100

0.000


M

0.38
0.29
0.39
0.25
0.33


IP VP IC
Validity


SD

0.03
0.03
0.06
0.05
0.02


M

0.48
0.30
0.48
0.46
0.39


SD

0.10
0.06
0.12
0.16
0.07


- LA
-*-HA


VC N


Figure 4.4. Group by validity interaction for saccade latency.


-:::










Simple effects analysis of the Validity x Session interaction (see Figure 4.5)

indicated a significant increase in saccade latency for IC from Session 1 (M = .391, SD =

.09) to Session 2 (M = .484, SD = .021) as well as a significant increase for VC from

Session 1 (M= .332, SD= .015) to Session 2 (M= .374, SD= .029). For the rest of the

cue validity types, there were non-significant changes. Thus, it appears that the

competitive session altered the central cue effect, but had little influence on the peripheral

cues.


0 .5 5 0


0 .5 0 0


0.4 5 0 P
--VP
0 .400 0 E
VC
0 .350 N


0 .3 0 0


0 .2 5 0
1 2
Session



Figure 4.5. Session by validity interaction for saccade latency.

ERP Data Analysis

After the data reduction was completed for the ERP data, a peak analysis

(Hoormann, Falkenstein, Schwarzenau, & Hohnsbein, 1998) was conducted for the P3 by

creating a search window. The search window was determined by computing the grand

means of the ERP (averages for one condition across all subjects) and the average of the










grand means across all conditions. The grand mean ERP is an estimate of the general

component structure of the ERP across conditions. Based on the grand mean ERP, an

ERP segment of 250 ms to 550 ms containing the P3 component was established.

Following the establishment of the search window, P3 was defined as the most positive

peak within search window, and then was obtained for each subject.

P3 amplitude was analyzed in a 2 (Group) x 3 (Site: Fz, Cz and Pz) x 2 (Validity)

x 2 (Session) mixed model ANOVA with repeated measures on the last factor (see Table

4.4 for means and standard deviations). Significant main effects were found for Site

(F(2, 52) = 68.32, p < .001), and for Session (F(1, 26) = 42.67, p < .001). However, more

important was the significant Site x Session interaction (F(2, 52) = 3.33, p < .05).

Table 4.4

P3 Amplitude



Session 1 Session 2
Site M SD M SD

CZ 8.63 2.90 6.06 2.43
FZ 4.70 2.61 4.24 2.34
PZ 9.91 2.51 7.54 2.13



12
10

> 0 Session 1
E 0 Session 2
4-
2
0
HA LA HA LA HA LA
CZ FZ PZ
Site by Group


Figure 4.6. P3 amplitude for the three sites across each session.











A simple effects test for the interaction (see figure 4.6) revealed a significant


reduction of P3 amplitude for CZ and PZ from Session 1 to Session 2, but little change


for FZ. Demonstrated was an overall reduction in P3 amplitude for CZ and PZ,


irrespective of trait anxiety level. Thus, it can be inferred that participation in the


competitive session (Session 2) decreased attentional allocation to the secondary task.


Curiously, the validity of the cue did not affect the P3 amplitude, as might be expected


(see figure 4.7).



10
9
8
7
6-

4-
3-
2-
1
0
2 r 2 ro 2 2 ro
> > >>
0 0 0
CZ FZ PZ
Site


Figure 4.7. P3 amplitude for site by amplitude.














CHAPTER 5
DISCUSSION, SUMMARY, CONCLUSIONS,
AND SUGGESTIONS FOR FUTURE RESEARCH

This study represented an attempt to specifically examine the central tenets of the

processing efficiency theory, and to further advance understanding of visual attention and

cortical activation in a simulated sport context. Based on the predictions of the PET,

several questions related to the proposed underlying mechanisms of anxiety, attention,

and arousal were addressed from theoretically based hypotheses. Specifically, the first

objective of this study was to determine whether higher anxious individuals, when

compared to lower anxious individuals, would demonstrate a reduction of resources for a

secondary task, an increase in resources for the primary task, and a reduction of

processing efficiency under a condition of high cognitive anxiety as based on the

dependent measures of driving speed, response time to peripheral lights, visual search

patterns, and mental effort assessed by ERP brain potentials. The second objective was

to record and evaluate cortical and visual search changes as they relate to human

performance. In particular, an attempt was made to determine the effects of attention

allocation in a highly stressful and information rich environment.

Initially in this chapter, the hypotheses presented in the introduction are discussed

in detail. Included is an elaboration and explanation of the direction of the results as they

pertain to the stated hypotheses and related theory. The next section represents a general

summary and subsequent conclusions concerning the processing efficiency theory and









attention allocation as they relate to the results. Finally, practical applications and a

discussion of future research directions are discussed.

Discussion

Cognitive Anxiety and Physiological Arousal

Two hypotheses were offered to address changes in cognitive anxiety and

physiological arousal. First, the anxiety and ego-threatening instructional set was

expected to produce higher levels of cognitive anxiety as measured by the CSAI-2 in the

competition session as compared to the baseline session. Furthermore, it was predicted

that the competitive session would produce higher cognitive anxiety for the HA when

compared to the LA group. These hypotheses were supported in that cognitive anxiety

increased for both groups from the baseline session (Session 1) to the competitive session

(Session 2). Furthermore, there was a larger increase in cognitive anxiety for the HA

group than the LA group. These results are supported by previous research, in which

attempts were made to increase anxiety through contrived instructional sets (e.g., Hardy,

1996; Hardy et al., 1994; Janelle et al., 1999, Williams & Davids, 1999). Remarkably,

the cognitive anxiety for the HA group was higher than has often been found in

laboratory settings. One possible explanation is the inclusion of the ego-threatening

portion of the instructional set, whereby participants were told that the competitive

session would be used to determine their motor skill ability, and that doing poorly would

demonstrate a lack of driving and motor skill ability. Based on the data, it seems that the

higher trait anxious participants viewed this to be more threatening than did the lower

trait anxious participants, thus invoking a comparatively greater increase in cognitive

anxiety.









Physiological arousal was evaluated through heart rate, which currently is a

widespread practice in sport anxiety research (e.g., Hardy, 1996; Hardy et al., 1994;

Hardy & Pates, 1995, Janelle et al., 1999). It was hypothesized that increases in

cognitive anxiety would produce a corresponding increase in heart rate in both groups.

Both groups demonstrated an increase in heart rate for the competitive session when

compared to the baseline session. There was not, however, a larger increase for higher

trait anxious participants, as was seen with cognitive anxiety. In research examining

heart rate change, large increases are usually not observed unless induced by contrived

physical activity that is not relevant to the task (e.g., shuttle runs before bowling; Hardy

et al., 1994). In this study, and as in Janelle et al. (1999), heart rate change was induced

by the anxiety and ego-threatening instructional sets. Thus, observed changes occurred

through a simple increase in cognitive anxiety. Though admittedly small, these increases

support the notion that heart rate reflects changes in arousal, and that increases in

cognitive anxiety will also increase physiological arousal.

Dual-task Performance

Eysenck and Calvo (1992) proposed that anxiety will produce an increase in effort

to improve or maintain performance of the current task, and that this effort will be larger

for highly anxious individuals. Furthermore, they emphasized the importance of

measuring performance effectiveness and processing efficiency separately due to the

notion that anxiety impairs processing efficiency to a greater extent than it impairs

performance effectiveness for higher trait anxious individuals relative to lower trait

anxious individuals. Thus, it was expected that the groups of interest would only slightly

differ on performance of the primary task, while significantly differing on performance of









the secondary task. Specifically, higher trait anxious individuals were predicted to have

significantly longer response times than lower trait anxious individuals to the secondary

task (e.g., Dibartolo, Brown, & Barlow, 1997) as a result of increased allocation of

attentional resources to maintain performance of the primary task. This hypothesis was

clearly supported. There was very little difference between the HA group and LA group

in the primary task. However, an overall increase in response time was evident for the

HA group as opposed to an overall reduction in response time for the LA group. The

introduction of anxiety caused an increase in attentional allocation to the driving task for

the HA group, thereby reducing the available resources for the secondary task.

Collectively, these results support the PET, and provide further evidence for the

performance effectiveness and processing efficiency distinction (e.g., Dibartolo et al.,

1997; Elliman et al., 1997).

The Cost and Benefit Paradigm

A key assumption of the PET is that attention is a limited resource. As

demonstrated by changes in primary and secondary task performance, the HA group

exhibited a reduction of resources for the secondary task. The secondary task was

presented as a probe task based on a cost and benefit paradigm. The goal was not only to

measure processing efficiency, but also to gather information concerning how this change

in resource allocation would influence cue identification and discrimination. It was

expected that reductions in processing resources for the secondary task would also result

in changes in visual attentional allocation. These changes could theoretically be

evidenced by the changes in the related cost of anticipating incorrectly or by the benefit

of anticipating correctly.









In the single task trial and non-competitive session, it was hypothesized that valid

peripheral cues would produce the fastest response times followed by valid central cues,

invalid central cues, and finally invalid peripheral cues which would produce the slowest

response times. Centrally located abrupt onset cues are believed to capture attention

voluntarily whereas the peripherally located abrupt onset cues are speculated to capture

attention in an involuntary, stimulus-driven manner (Theeuwes, et al., 1998). Partial

support for this hypothesis was found for the single task trial. Cue validity was affected

by response time, but location of the cue was not. It was found that both invalid cues

induced a cost and both valid cues produced a benefit. The non-competitive session and

the competitive session produced the longest response times for the invalid peripheral

cues followed by invalid central cues, and valid central cues, with valid peripheral cues

producing the fastest response times. Jonides (1981) originally found that cue location

played an important role in attention allocation. This was not replicated in the single

trial, which could be potentially due to methodological differences. In Jonides study,

participants were biased by instruction or involvement in a dual-task situation, but in the

current single task trial participants were only told to perform the task. The dual-task

session required attention to be divided and resulted in participants being more vulnerable

to the abrupt onset of peripherally located cues.

The second hypothesis predicted changes in the central cue effect for the HA

group. Also proposed was that due to the conscious and voluntary nature of central cues,

they would be more susceptible to the reduction of resources. Therefore, it was expected

that response times would increase for valid centrally located cues in the competitive

session for high trait anxious individuals when compared to the non-competitive session.









Also hypothesized was that valid and invalid peripheral cues for both groups were

expected to be unaffected by the decrease in processing space for the secondary task. As

stated earlier, peripheral cues were predicted to be resistant to increased demands on

processing capacity. Thus, changes in resource allocation would not interfere with the

processing of peripheral cues. These hypotheses were not supported. Because there were

increases in response time for all cues between Session 1 and Session 2 for the HA group,

claiming that central cues are different from peripheral cues in their level of automaticity,

at least in the context of this investigation, is not warranted. The Group by Session

interaction represented an increase in response time for all cue types for the HA group

and a reduction of all cue types for the LA group. In part, the results presented in

Jonides' study (1981) were not substantiated. The peripheral cues were not resistant to

distraction of anxiety on the central task, and the results did not demonstrate the expected

automaticity effect as suggested by Jonides (1981). Furthermore, it does not appear that

the central cue effect was altered by the introduction of anxiety.

The response time results and lack of support for some of Jonides' conclusions

increases the support for the PET two fold. First, the human brain has a limited capacity

for processing information, and increases in anxiety result in a reduction of the working

memory capacity. In the PET, the presence of anxiety causes a shift in processing

resources, which is most readily measured in a dual task paradigm (as was the case here).

The data signify an increase in attention allocation to the primary task as induced by the

control unit for higher anxious individuals, thus an overall increase of response time for

the secondary task. Second, and conversely, lower anxiety allows for more processing

space, thus increasing the total space available for multiple tasks. Lower anxious









individuals were not comprised as much by the presence of anxiety. Thus, they had more

processing space available to perform both primary and secondary tasks well.

While there is ample evidence to demonstrate that peripheral cues capture

attention in an involuntary stimulus-driven manner (e.g., Jonides, 1981, Remington et al.,

1992, Theeuwes et al., 1998), the results in these studies are often derived from less

dynamic and stress invoking situations than was the case in this study. Likewise,

previous results have been derived in environments that are void of anxiety. The notion

that peripheral abrupt onsets drive attention involuntary has been argued to represent an

innate reaction to detect and respond to events that may require immediate action

(Theeuwes et al., 1998).

While acknowledging that attentional processing typically proceeds in a manner

consistent with the findings of Theeuwes and colleagues, with the introduction of anxiety

one might consider the concept of attentional narrowing (Easterbrook, 1959) as a

plausible explanation for the current findings. Easterbrook's cue utilization theory

proposes that during dual-task situations, the impact of arousal on performance will

change dependent on the current arousal level. The narrowing of attention, due to the

increase of arousal, results in peripheral (or secondary task) information being filtered

out. The logic follows that, at moderate levels, irrelevant or distracting cues are filtered

from processing, but that as arousal continues to increase, processing of both irrelevant

and relevant cues are compromised. The result is a funneling of attention and a reduction

in scope of the attentional focus. A common metaphor partially based on Easterbrook's

ideas, equates the focus of attention with the beam of a 'spotlight' (Castiello & Umilta,

1990). When arousal is low, the 'spotlight' is broad, but as arousal increases the




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs